Handbook of Fish Biology and Fisheries VOLUME 2 FISHERIES

EDITED BY Paul J.B. Hart Department of Biology University of Leicester AND John D. Reynolds School of Biological Sciences University of East Anglia

HANDBOOK OF FISH BIOLOGY AND FISHERIES Volume 2 Also available from Blackwell Publishing:

Handbook of Fish Biology and Fisheries Edited by Paul J.B. Hart and John D. Reynolds Volume 1 Fish Biology Handbook of Fish Biology and Fisheries VOLUME 2 FISHERIES

EDITED BY Paul J.B. Hart Department of Biology University of Leicester AND John D. Reynolds School of Biological Sciences University of East Anglia ©ȱ2002ȱbyȱBlackwellȱScienceȱLtdȱ aȱBlackwellȱPublishingȱcompanyȱ ȱ Chapterȱ8ȱ©ȱBritishȱCrownȱcopyright,ȱ1999ȱ ȱ BLACKWELLȱPUBLISHINGȱ 350ȱMainȱStreet,ȱMalden,ȱMAȱ02148Ȭ5020,ȱUSAȱ 108ȱCowleyȱRoad,ȱOxfordȱOX4ȱ1JF,ȱUKȱ 550ȱSwanstonȱStreet,ȱCarlton,ȱVictoriaȱ3053,ȱAustraliaȱ ȱ TheȱrightȱofȱPaulȱJ.B.ȱHartȱandȱJohnȱD.ȱReynoldsȱtoȱbeȱidentifiedȱasȱtheȱAuthorsȱȱ ofȱtheȱEditorialȱMaterialȱinȱthisȱWorkȱhasȱbeenȱassertedȱinȱaccordanceȱwithȱtheȱȱ UKȱCopyright,ȱDesigns,ȱandȱPatentsȱActȱ1988.ȱ ȱ Allȱrightsȱreserved.ȱNoȱpartȱofȱthisȱpublicationȱmayȱbeȱreproduced,ȱstoredȱinȱaȱretrievalȱ system,ȱorȱtransmitted,ȱinȱanyȱformȱorȱbyȱanyȱmeans,ȱelectronic,ȱmechanical,ȱ photocopying,ȱrecordingȱorȱotherwise,ȱexceptȱasȱpermittedȱbyȱtheȱUKȱCopyright,ȱ Designs,ȱandȱPatentsȱActȱ1988,ȱwithoutȱtheȱpriorȱpermissionȱofȱtheȱpublisher.ȱ ȱ Firstȱpublishedȱ2002ȱ Reprintedȱ2004ȱ ȱ LibraryȱofȱCongressȱCatalogingȬinȬPublicationȱDataȱhasȱbeenȱappliedȱfor.ȱ ȱ Volumeȱ1ȱISBNȱ0Ȭ632Ȭ05412Ȭ3ȱ(hbk)ȱ Volumeȱ2ȱISBNȱ0Ȭ632Ȭ06482ȬXȱ(hbk)ȱ 2ȬvolumeȱsetȱISBNȱ0Ȭ632Ȭ06483Ȭ8ȱ ȱ AȱcatalogueȱrecordȱforȱthisȱtitleȱisȱavailableȱfromȱtheȱBritishȱLibrary.ȱ ȱ Setȱinȱ9/11.5ȱptȱTrumpȱMediaevalȱ byȱSNPȱBestȬsetȱTypesetterȱLtd,ȱHongȱKongȱ PrintedȱandȱboundȱinȱtheȱUnitedȱKingdomȱȱ byȱTJȱInternationalȱLtd,ȱPadstow,ȱCornwall.ȱ ȱ Theȱpublisher’sȱpolicyȱisȱtoȱuseȱpermanentȱpaperȱfromȱmillsȱthatȱoperateȱaȱsustainableȱ forestryȱpolicy,ȱandȱwhichȱhasȱbeenȱmanufacturedȱfromȱpulpȱprocessedȱusingȱacidȬfreeȱandȱ elementaryȱchlorineȬfreeȱpractices.ȱFurthermore,ȱtheȱpublisherȱensuresȱthatȱtheȱtextȱpaperȱ andȱcoverȱboardȱusedȱhaveȱmetȱacceptableȱenvironmentalȱaccreditationȱstandards. ȱ Forȱfurtherȱinformationȱonȱ BlackwellȱPublishing,ȱvisitȱourȱwebsite:ȱ http://www.blackwellpublishing.comȱ Contents

Contributors, x Preface, xii Abbreviations, xiv

1 THE HUMAN DIMENSION OF FISHERIES SCIENCE, 1 Paul J.B. Hart and John D. Reynolds 1.1 Introduction, 1 1.2 The origins of fisheries science, 1 1.3 The objectives of fisheries management, 2 1.4 The development of the institutions of fisheries science, 3 1.5 Who owns the fishery and for whom is it managed?, 4 1.6 The problem of uncertainty, 4 1.7 Development to increase economic efficiency or freedom? 6 1.8 Achieving sustainable development, 7 1.9 Conclusions, 8

Part 1: Background to Fisheries, 11

2 FISH CAPTURE DEVICES IN INDUSTRIAL AND ARTISANAL FISHERIES AND THEIR INFLUENCE ON MANAGEMENT, 13 Ole Arve Misund, Jeppe Kolding and Pierre Fréon 2.1 Introduction, 13 2.2 Main fish capture techniques, 13 2.3 Artisanal fisheries, 25 2.4 Conclusions, 31

3 MARKETING FISH, 37 J.A. Young and J.F. Muir 3.1 Introduction, 37 3.2 Marketing and markets, 37 vi Contents

3.3 Determination of values, 42 3.4 Creating values, 45 3.5 Communicating values, 48 3.6 Delivering values, 51 3.7 Future values, 53 3.8 Conclusions, 56

4 A HISTORY OF FISHERIES AND THEIR SCIENCE AND MANAGEMENT, 61 Tim D. Smith 4.1 The nature of fishing, 61 4.2 Origins of fishing, 63 4.3 Overfishing, 65 4.4 Graham’s Law of Fishing, 69 4.5 Scientific bases for management, 71 4.6 Post-Second World War, 75 4.7 Conclusions, 80

5 GATHERING DATA FOR RESOURCE MONITORING AND FISHERIES MANAGEMENT, 84 David Evans and Richard Grainger 5.1 Fisheries information framework, 84 5.2 Fisheries information data, 91 5.3 Fisheries data collection and management, 94 5.4 Fisheries information presentation, 99 5.5 Conclusions, 101

Part 2: Stock Assessment, 103

6 SURPLUS PRODUCTION MODELS, 105 Jon T. Schnute and Laura Richards 6.1 Introduction, 105 6.2 Graphical model, 107 6.3 Classical differential equations (model 1), 109 6.4 Special cases (model 1), 111 6.5 Equilibrium (model 1), 112 6.6 Growth models, 114 6.7 Modern difference equations (model 2), 116 6.8 Equilibrium and dynamic response (model 2), 118 6.9 Estimation: comparing models with data, 122 6.10 Model extensions, 123 6.11 Conclusions, 124 Contents vii

7 DYNAMIC POOL MODELS I: INTERPRETING THE PAST USING VIRTUAL POPULATION ANALYSIS, 127 J.G. Shepherd and J.G. Pope 7.1 Introduction: the dynamic pool concept, 127 7.2 The distinction between retrospective analysis and forecasting, 127 7.3 Virtual population analysis (VPA): the basics, 128 7.4 Separable VPA, 135 7.5 Tuning of VPA using CPUE and survey data, 141 7.6 The extended survivors method (XSA), 149 7.7 Multispecies virtual population analysis, 156 7.8 Conclusions, 161

8 DYNAMIC POOL MODELS II: SHORT-TERM AND LONG-TERM FORECASTS OF CATCH AND BIOMASS, 164 J.G. Shepherd and J.G. Pope 8.1 Short-term forecasts of catch and biomass, 164 8.2 Long-term forecasts of catch and biomass, 172 8.3 Multispecies forecasts, 184 8.4 Conclusions, 186

9 A BUMPY OLD ROAD: SIZE-BASED METHODS IN FISHERIES ASSESSMENT, 189 Tony J. Pitcher 9.1 Introduction, 189 9.2 Age and mortality methods, 190 9.3 When length is king, 204 9.4 Conclusions, 207

10 ECOSYSTEM MODELS, 211 Daniel Pauly and Villy Christensen 10.1 Introduction: ecosystem models, 211 10.2 Multispecies models, 213 10.3 Modelling the North Sea, the North Pacific and the Gulf of Thailand, 214 10.4 Ecopath and the mass-balance approach, 215 10.5 The trophic level concept, 217 10.6 Practical uses of trophic levels: tracking food-web changes, 219 10.7 Ecosystem structure, Odum’s maturity and Ulanowicz’s ascendancy, 220 10.8 Particle-size spectra in ecosystems, 221 10.9 Ecosim, refugia and top-down vs. bottom-up control, 222 10.10 Spatial considerations in ecosystem modelling, 223 10.11 Towards a transition from single-species to ecosystem-based management, 224 10.12 Conclusions, 225 viii Contents

11 INDIVIDUAL-BASED MODELS, 228 Geir Huse, Jarl Giske and Anne Gro Vea Salvanes 11.1 Introduction, 228 11.2 Specifying individuals in IBMs, 229 11.3 Features of individual-based models, 230 11.4 Formulating and testing IBMs, 236 11.5 Review of individual-based models in fisheries biology, 237 11.6 Conclusions, 244

12 THE ECONOMICS OF FISHERIES, 249 Rögnvaldur Hannesson 12.1 Introduction, 249 12.2 The surplus production model, 249 12.3 Fishing effort and fish yield, 250 12.4 Sustainable yield, 252 12.5 Optimal exploitation, 254 12.6 Time discounting, 257 12.7 Fluctuations: should catches be stabilized?, 260 12.8 Optimum fleet capacity for fluctuating stocks, 261 12.9 Management methods, 263 12.10 International issues, 265 12.11 Conclusions, 267

13 CHOOSING THE BEST MODEL FOR FISHERIES ASSESSMENT, 270 Per Sparre and Paul J.B. Hart 13.1 Introduction, 270 13.2 Basic concepts, 270 13.3 Stochastic modelling, 272 13.4 Mechanics of choosing a model, 274 13.5 Estimation of parameters, 275 13.6 The use of mathematical models in fisheries, 276 13.7 A key to models, 283 13.8 The importance of testing models, 286 13.9 Conclusions, 287

Part 3: Fisheries in a Wider Context, 291

14 MARINE PROTECTED AREAS, FISH AND FISHERIES, 293 Nicholas V.C. Polunin 14.1 Introduction, 293 14.2 What happens to target species in MPAs?, 294 14.3 What are the potential benefits to fishers?, 298 14.4 What else do MPAs offer?, 305 Contents ix

14.5 Realities, advocacy and implementation of MPAs, 307 14.6 Conclusions, 311

15 EXPLOITATION AND OTHER THREATS TO FISH CONSERVATION, 319 John D. Reynolds, Nicholas K. Dulvy and Callum M. Roberts 15.1 Introduction, 319 15.2 Global status of exploited fish populations, 321 15.3 Extinction, 321 15.4 Exploitation as a cause of declines and extinction, 325 15.5 What renders species susceptible to overfishing?, 329 15.6 Conservation meets sustainable use, 332 15.7 What is needed to safeguard fish biodiversity?, 334 15.8 Conclusions, 336

16 ECOSYSTEM EFFECTS OF FISHING, 342 M.J. Kaiser and S. Jennings 16.1 Introduction, 342 16.2 Effects of harvesting target species, 342 16.3 Effects of prey removal on sea birds, 347 16.4 How fisheries affect energy subsidies, 349 16.5 Effects of removing predators, 353 16.6 Effects on benthic communities and habitats, 354 16.7 Bycatches and ghost-fishing, 358 16.8 Conclusions, 361

17 RECREATIONAL FISHING, 367 I.G. Cowx 17.1 Introduction, 367 17.2 Types of recreational fishing, 368 17.3 Recreational fishery assessment methods, 373 17.4 Management of recreational fisheries, 376 17.5 Issues relating to the development of recreational fisheries, 380 17.6 Value, 385 17.7 Conclusions, 385

Index, 391

Colour plate falls between p. 194 and p. 195 Contributors

Villy Christensen Fisheries Centre, Universi- Rögnvaldur Hannesson Norwegian School ty of British Columbia, 2204 Main Mall, of Economics and Business Administration, Vancouver, BC V6T 1Z4, Canada Helleveien 30, N-5045 Bergen, Norway v.christensen@fisheries.ubc.ca [email protected]

I.G. Cowx Hull International Fisheries Insti- Paul J.B. Hart Department of Biology, Univer- tute, Department of Biological Sciences, Univer- sity of Leicester, Leicester LE1 7RH, UK sity of Hull, Hull HU6 7RX, UK [email protected] [email protected] Geir Huse Department of Fisheries and Marine Nicholas K. Dulvy Department of Marine Sci- Biology, University of Bergen, PO Box 7800, ences and Coastal Management, University of N-5020 Bergen, Norway Newcastle, Newcastle upon Tyne, NE1 7RU, UK [email protected] [email protected] S. Jennings The Centre for Environment, David Evans Fisheries Adviser, 69, Kingsmead Fisheries and Aquaculture Science, Lowestoft Road, London, SW2 3HZ, UK Laboratory, Pakefield Road, Lowestoft, Suffolk evans.fi[email protected] NR33 0HT, UK [email protected] Pierre Fréon Marine and Coastal Manage- ment, Private Bag X2, Rogge Bay 8012, Cape Town, South Africa M.J. Kaiser School of Ocean Sciences, Univer- [email protected] sity of Wales – Bangor, Menai Bridge, Anglesey, LL59 5EY, UK Jarl Giske Department of Fisheries and [email protected] Marine Biology, University of Bergen, PO Box 7800, N-5020 Bergen, Norway Jeppe Kolding Department of Fisheries and [email protected] Marine Biology, University of Bergen, PO Box 7800, N-5020 Bergen, Norway Richard Grainger Fisheries Department, [email protected] Food and Agricultural Organization of the United Nations, Viale delle Terme di Caracalla, 00100 Ole Arve Misund Institute of Marine Re- Rome, Italy search, PO Box 1870, N-5817 Bergen, Norway [email protected] [email protected] Contributors xi

J.F. Muir Institute of Aquaculture, University Callum M. Roberts Environment Depart- of Stirling, Stirling FK9 4LA, UK ment, University of York, York, YO10 5DD, UK [email protected] [email protected]

Daniel Pauly Fisheries Centre, University of Anne Gro Vea Salvanes Department of Fish- British Columbia, 2204 Main Mall, Vancouver, eries and Marine Biology, University of Bergen, BC V6T 1Z4, Canada PO Box 7800, N-5020 Bergen, Norway d.pauly@fisheries.ubc.ca [email protected]

Tony J. Pitcher Fisheries Centre, University of Jon T. Schnute Department of Fisheries British Columbia, 2204 Main Mall, Vancouver, and Oceans, Pacific Biological Station, 3190 BC V6T 1Z4, Canada Hammond Bay Road, Nanaimo, BC V9R 5K6, t.pitcher@fisheries.ubc.ca Canada [email protected] Nicholas V.C. Polunin Department of Ma- rine Sciences and Coastal Management, Uni- J.G. Shepherd School of Ocean and Earth Sci- versity of Newcastle, Newcastle upon Tyne, NE1 ence, University of Southampton, Southampton 7RU, UK Oceanography Centre, European Way, [email protected] Southampton, SO14 3ZH, UK [email protected] J.G. Pope The Norwegian College of Fishery Science, University of Tromsø, N-9037 Tromsø, Tim D. Smith Northeast Fisheries Science Norway Center, 166 Water Street, Woods Hole, MA 02543- [email protected] 1026, USA [email protected] [email protected]

John D. Reynolds School of Biological Sci- Per Sparre Danish Institute for Fisheries ences, University of East Anglia, Norwich NR4 Research, Department of Marine Fisheries, 7TJ, UK Charlottenlund Slot, DK-2920 Charlottenlund, [email protected] Denmark [email protected] Laura Richards Department of Fisheries and Oceans, Pacific Biological Station, 3190 J.A. Young Department of Marketing, Univer- Hammond Bay Road, Nanaimo, BC V9R 5K6, sity of Stirling, Stirling FK9 4LA, UK Canada [email protected] [email protected] Preface

The goal of the two volumes of the Handbook of coverage of fish biology simply because the topics Fish Biology and Fisheries is to help integrate the are interesting in their own right, and because study of fish biology with the study of fisheries. the borders between pure and applied research are One might not expect these two subjects to need fuzzy. We therefore decided not to restrict infor- further integration. However, strong declines in mation according to its direct relevance to fish- many fish stocks around the globe, combined with eries as this subject is understood today. We hope growing concerns about the impact of fisheries the result will be of value to undergraduates and on marine and freshwater biodiversity, are raising graduates looking for information on a wide vari- new questions about aspects of fish biology that ety of topics in fisheries science. The books are also have traditionally dwelt outside mainstream fish- aimed at researchers who need up-to-date reviews eries research. Thus, fisheries biologists and of topics that impinge on their research field but managers are increasingly asking about aspects of may not be central to it. The information should ecology, behaviour, evolution and biodiversity also be useful to managers and decision makers that had traditionally been studied by different who need to appreciate the scientific background people who attend different conferences and pub- to the resources they are trying to manage and lish in different journals. By bringing these people conserve. and their subjects together in the two volumes of In the first volume, subtitled Fish Biology, our this Handbook, we hope to foster a better two-way introductory chapter explores the underpinnings flow of information between the studies of fish of fisheries biology and management by basic biology and fisheries. research on fish biology. Part 1 then examines A tradition runs through the prefaces of the systematics and biogeography of fishes, including other volumes in this series whereby the editors methods for determining phylogenetic relation- distance themselves from a literal translation of ships and understanding spatial patterns of diver- the word ‘Handbook’. In keeping with this tradi- sity. Part 2 examines production and population tion, we wish to make clear that this is not a cook- structures of fishes, beginning with chapters on book of recipes for how to study and manage fish physiology and growth, followed by recruitment, populations. Instead, we have tried to produce a life histories, migration, population structure and pair of reference books that summarize what is reproductive ecology. Part 3 considers fishes as known about fish biology and ecology, much of predators and as prey, making use of conceptual which is relevant to assessment and management advances in behavioural ecology to link predator– of fish populations and ecosystems. Of course, prey interactions to the environment. In Part much of the material in the first volume may never 4 we scale up from individual interactions to find applications in fisheries, and that is fine with communities and ecosystems. The chapters us. We encouraged our authors to provide a wide include comparisons of freshwater and marine Preface xiii communities, as well as interactions between The final stages of production have been greatly fishes and parasites. helped by two people. Valery Rose of Longworth In the second volume, subtitled Fisheries, we Editorial Services worked patiently and efficiently begin with a chapter that considers the human to shepherd the book through the copy editing dimension of fisheries management. Part 1 then stages, and Monica Trigg, did an excellent job with gives background information for fisheries, in- the gargantuan task of constructing the two in- cluding fishing technology, marketing, history of dices. Paul Hart would like to thank the Uni- fisheries, and methods of collecting and present- versity of Bergen, Department of Fisheries and ing data. Part 2 provides fundamental methods of Marine Biology, for providing him with space and stock assessment, including surplus production support during the final editing of the manuscript. models, virtual population analyses, methods for John Reynolds would like to thank his many col- forecasting, length-based assessments, individual- leagues at the University of East Anglia and at the based models and economics. We have also tried Centre for Environment, Fisheries, and Aquacul- to consolidate this information by reviewing the ture Science (Lowestoft) who have helped build various options available for modelling fisheries, bridges between pure and applied research. including the pros and cons of each. Part 3 explores wider issues in fisheries biology and management, The editors and publisher would like to thank including the use of marine protected areas, con- copyright-holders for permission to reproduce servation threats to fishes, ecosystem impacts and copyright material. They apologize for any errors recreational fishing. or omissions, and would be grateful to be notified We are very grateful to our 54 authors from 10 of any corrections that should be incorporated in countries for their hard work and patience while the next edition or reprint of this book. we attempted to herd them all in the same general direction. We also thank Susan Sternberg who first Paul J.B. Hart and John D. Reynolds suggested that we take on this project, and Delia Leicester and Norwich Sandford at Blackwell Sicence who has overseen it. Abbreviations

ANN artificial neural network MSC Marine Stewardship Council BPR biomass per recruit MSVPA multispecies virtual population CPUE catch per unit of effort analysis EEZ Exclusive Economic Zone MSY maximum sustainable yield ELEFAN electronic length–frequency NAFO North Atlantic Fisheries analysis Organization FAO Food and Agriculture Organization SDP stochastic dynamic programming GOF goodness of fit SFIA Sea Fish Industry Authority IBM individual based models SLCA Shepherd’s length–composition ICES International Council for the analysis Exploration of the Sea SRR stock–recruitment relationship ICLARM International Center for Living SSB spawning stock biomass Aquatic Resources Management TAC total allowable catch ITQ individual transferable quota UNCLOS United Nations Convention on the JAM judicious averaging method Law of the Sea MBA Mariue Biological Association USDA US Department of Agriculture MEY maximum economic yield XSA extended survivors analysis MIX statistical mixture analysis VPA virtual population analysis MPA marine protected area YPR yield per recruit 1 The Human Dimension of Fisheries Science

PAUL J.B. HART AND JOHN D. REYNOLDS

1.1 INTRODUCTION stock and how their abundance is likely to change in the short to medium term. In this context Fisheries science and fisheries management are they discuss reference points for management. not the same thing. Fisheries science is a multidis- Hannesson’s chapter (Chapter 12) deals extensively ciplinary subject that integrates animal behaviour, with fisheries management methods, particularly ecology and population dynamics with environ- from an economic point of view. Clearly, econom- mental processes to predict how animal popu- ics plays a key role in every aspect of commercial lations respond to fishing mortality. The results fisheries, from the intensity of fishing and kinds of of fisheries science inform fisheries manage- gear used, to the short-term costs and long-term ment, whereby policies are implemented to meet benefits to be derived from conserving stocks. specific objectives set by various stakeholders Young and Muir (Chapter 3) describe the attempt ranging from fishers to consumers to conserva- by the Marine Stewardship Council to employ tionists. The focus on science in the two volumes accreditation of fisheries and labelling of their of this Handbook reflects our own backgrounds products to use consumer pressure to improve the but is also a consequence of the way in which fish- standard of fisheries management. This chapter ery science and management have developed. In also points out that the fishing industry is almost this chapter we try to outline some of the fisheries entirely supply driven with little regard for what management issues that are not dealt with else- the consumer wants. Finally Evans and Grainger where in any detail, but that need to be considered (Chapter 5) outline how management policy is to put into context the state of fisheries science. integrated with data collection for fishery moni- Most of the chapters in this second volume con- toring and assessment. centrate on the technical issues of either assessing a fish stock or assessing the effects that the fishery has on the ecosystem – in other words, fishery 1.2 THE ORIGINS OF science (Hutchings et al. 1997a). However, various FISHERIES SCIENCE aspects of management are found throughout this book. For example, Misund et al. (Chapter 2) Biologists, or more properly zoologists, were the discuss the interaction between fisheries manage- first to draw attention to the problem of overfish- ment and the type of gear used, particularly in ing (Smith, Chapter 4, this volume). According to artisanal fisheries. Gear types dictate the kinds Graham (1956), fisheries science took its charac- of management measures that can be taken. teristic form from around 1890 onwards with a Shepherd and Pope (Chapters 7 and 8) provide blend of zoology and statistics but each with a new methods for determining how many fish are in a form and function: ‘The form was knowledge like 2 Chapter 1 that of a fisherman, and function was guidance to are simple on the surface but have wider implica- better use of the stocks of fish.’ This foreshadows tions. Off the south coast of Devon, UK, the in- the division already pointed to between fisheries shore waters around Start Point midway between science and management. In the same book, Plymouth in the west and Brixham in the east are Beverton and Holt (1956) wrote that fisheries man- partitioned (Hart 1998). Certain areas are closed agement should aim to adjust the fishing activity to all mobile gears such as trawlers or scallop to obtain the best results. They defined ‘best’ as dredgers and are intended for the exclusive use of being ‘the greatest sustained yield’. The search for crab potters. Other areas are open to mobile gear this single, biologically based objective for fish- for parts of the year when crab fishing is at its eries management came to dominate fisheries sci- seasonal nadir. The principal objective of this ence. In the early days most biologists seemed to agreement is to keep mobile and fixed gear apart. operate on the principle that if they could make Otherwise the crab potters would constantly be sure that the supply of fish could be sustained, the having their gear destroyed by trawlers or scallop commercial side would look after itself. This is dredgers. Although this is the principal objective, reminiscent of the views of some fisheries biolo- there are wider consequences both for the fishing gists today when confronted with questions about community and for the biological community. biodiversity: if the supply of fish is maintained, Without the agreement the potters would not be extinction risk will be minimal (reviewed by able to make a living. As there is little alternative Reynolds et al., Chapter 15, this volume). employment, many people would have to move away, destroying the structure of small local villages such as Beesands and Hallsands and 1.3 THE OBJECTIVES OF removing secondary jobs from the wholesale and FISHERIES MANAGEMENT transport business. From a biological perspective, the areas closed to mobile gear meet another objec- It is now recognized that there are many reasons tive by enabling a more complex benthic commu- for managing fisheries and that most of them are nity, which includes larger individuals than in the not to do with yield maximization. The issue of areas that are trawled regularly (Kaiser et al. 2000, the objectives of fisheries management is critical. Kaiser and Jennings, Chapter 16, this volume). Al- The early view that we should aim to sustain the though the conservation of biodiversity was not an highest catch possible no longer holds. It is not suf- original objective of the partitioning system, this ficiently precautionary (Punt and Smith 2001) and has been a fortunate by-product. it ignores other objectives. Hilborn and Walters The past preoccupation with maximizing (1992) organize the objectives into four categories: sustainable yield was thought sufficient for many (1) biological, (2) economic, (3) recreational and years even though fisheries economists were (4) social. In most cases the objectives for a par- already commenting in the 1950s that fisheries ticular management regime will be a mixture of management was not about biology but about the four. For example, the Norwegian government fishers’ behaviour (Smith, Chapter 4, this volume). subsidizes small communities in the north to keep The same point was made by Wilen (1979) but this people living in the region and one of the main thinking has not penetrated many of the institu- economic activities is fishing. It would not do, tions that assess exploited stocks (but see Hilborn though, to wreck this resource base by fishing et al. 1993). A focus restricted to the biological unsustainably. So we have two objectives work- aspects of fisheries science did not grow to fully ing in parallel: to maintain the communities by embrace the wider issues of economics and social supporting their infrastructure at the taxpayers’ structure. Why not? expense and to manage the fisheries sustainably. The objectives of some management systems The Human Dimension of Fisheries Science 3

industry, who see scientists as at best a bunch of 1.4 THE DEVELOPMENT OF boffins and at worst as agents of the government. THE INSTITUTIONS OF This put distance between the people who assess FISHERIES SCIENCE fish stocks and those exploiting the stocks. Once this gulf had developed it was easy for each side to Part of the answer must come from the way in argue that the other did not really appreciate what which fisheries science has become institu- the problem was. Today, attempts to bridge this tionalized. In Great Britain those who first drew gulf are an important element in the drive to create attention to the problems of overfishing were often suitable management measures. employed most of the time elsewhere. One of the Although early workers appreciated that eco- most famous was Thomas Henry Huxley, who had nomic questions were important to understanding wide interests in science and earned much of his fisheries (Smith, Chapter 4, this volume), it some- income from writing and from taking various jobs how took a long time for people with expertise such as Inspector for Salmon Fisheries (Lankester in economics or sociology to be employed by the 1895; Clark 1968). Walter Garstang was primarily institutions that assessed the stocks. Even now, in- an academic zoologist who interested himself in stitutions such as the International Council for the fisheries question only as a sideline. The activ- the Exploration of the Sea (ICES) have little input ities of these people led eventually to the establish- from economists and sociologists. Fish stock ment of bodies such as the Marine Biological assessment has usually stopped at the assessment Association of the UK (MBA) whose original pur- of the biological productivity of the stock when pose was: ‘To establish and maintain laboratories really it would be a good idea to also have people on the coast of the United Kingdom, where accu- who could evaluate the economic or social bene- rate researches may be carried on, leading to im- fits of proposed management policies. This point is provement of zoological and botanical science, and taken up again later when we discuss methods for to an increase of our knowledge as regards the food, dealing with uncertainty. life-conditions, and habits of British food-fishes Within fisheries laboratories the emphasis is and molluscs’ (Lankester 1895; Lee 1992). Early often on scientific research. Assessment work is workers employed by the MBA, such as Ernest regarded as routine and not conducive to career Holt, worked very closely with the fishing indus- advancement (Finlayson 1994). While fundamental try and spent time on trawlers and at the fish mar- research that leads to high-quality refereed publi- ket sampling fish. As institutions became larger cations is obviously very satisfying and important and employed more people, they evolved their own in many contexts, institutions with stock assess- ethos and traditions. This led to subtle changes in ment responsibilities need to develop career ap- the objectives of employees. In the end the MBA, praisal criteria which reward people for doing good for example, became a much more academic stock assessments and generating sound advice. research institution focusing on the ‘zoological There is a parallel here with engineering. When and botanical science’ part of its brief, while its designers and engineers set to work to construct fisheries function was taken over by the nascent the bridge between Malmö (Sweden) and Copen- Fisheries Laboratory at Lowestoft, now called the hagen (Denmark) over the Öresund they would Centre for Environment, Fisheries and Aquacul- have depended heavily on a wide range of scientific ture Science (Graham 1956; Lee 1992). laws and principles. But no one would have been Once established as institutions charged with happy if the bridge engineers were rewarded more carrying out stock assessments and giving advice for a general theory of bridge collapse than for to politicians about implications of management choosing the correct equations to tell them how measures, government-operated fisheries labora- thick the suspension wires should be. tories were easy targets for people in the fishing 4 Chapter 1

So far we have considered the fisher as the sole 1.5 WHO OWNS THE stakeholder, although perhaps operating on behalf FISHERY AND FOR WHOM of society. What do we mean by ‘society’? Just as IS IT MANAGED? different fishers often have competing perspec- tives, different members of society have different The separation of fishery scientists and managers views. Consumers want to be able to buy fish to into institutions, and their alienation from the eat, and in many developing countries they have fishing industry, has resulted in the two groups few or no dietary alternatives. Members of society having very different objectives. Fishery scientists are also increasingly concerned about biodiver- want to control and regulate the industry while sity and ecosystem structure in aquatic habitats fishers want to earn at least enough money to pay (Reynolds et al., Chapter 15, this volume). The their costs and make a living. Who are fisheries sci- entry of conservationists into the sustainable use entists working for? If we take the view that they dialogue has put new demands on fisheries science are working for fishers, perhaps we would expect and management (Reynolds et al. 2001). Whatever their perspectives to coincide. But of course this is rights fishers may have to make a living from the unrealistic, because fishers compete with other sea, conservationists argue that these rights do not fishers, and scientists are often asked for advice extend to the decimation of populations for profit. that can be used to referee the competition. This According to this viewpoint, if fishers cannot meet competition often takes on nationalistic fervour: conservation objectives, they should stop fishing. ‘Why should we struggle to make ends meet under reduced quotas when fishers from another country take our fish?’ Another view is that society owns the fishery, 1.6 THE PROBLEM and fishers are given permission to exploit the re- OF UNCERTAINTY source and should be grateful for this chance. In a legal sense fish in coastal waters are indeed typi- We began Chapter 1 in Volume 1 with Charles’s cally the property of the state (Charles 1998a; (1998b) categorization of uncertainty in fisheries Walters 1998). From this Walters (1998) argues that science. These were (i) random fluctuations, (ii) fishers have no rights, and that exploitation should uncertainty in parameters and states of nature, and only occur if this benefits society. The same (iii) structural uncertainty. It is becoming increas- argument is made about forestry companies when ingly evident that one of the biggest challenges for conservationists question the rights of loggers to the future is to devise institutional systems that exploit a common resource. When we contrast this are able to take account of uncertainty and to take view with the fishers’ viewpoint, we see a recipe decisions in the face of it. A number of fisheries for failure of many standard management mea- have collapsed, probably because uncertainty about sures. The fisher borrows money from the bank to their state was either ignored or played down. buy a boat and has to catch enough fish to pay off We outline two examples. his loan and to maintain his crew and his family Between 1920 and 1945 the California sardine during the life of the boat. Taking account of the (Sardinops sagax) formed the basis of an impor- fisher’s perspective therefore requires that manage- tant fishmeal and canning industry centred on ment measures manipulate the costs and benefits Monterey, USA (McEvoy 1986; Mangelsdorf 1986). that direct the actions of the fisher. Regulations Factory owners soon realized that converting the based on total allowable catches and minimum whole fish into meal and abandoning the canning landing sizes will be forever compromised because process could earn them more money. As the fish- they require fishers to throw away money (dead fish) ery increased, evidence for impacts on the stock for the good of society or their competitors. Few was evaluated by both the California Fish and individuals in any walk of life are happy doing that. Game (CFG) Fisheries Laboratory and the US The Human Dimension of Fisheries Science 5

Bureau of Fisheries’ Laboratory (USBF). The The collapse of the cod fishery off Newfound- former was dedicated to the conservation of land is our second example. We will keep the Californian resources and one of the measures details to a minimum because much has already taken to control fishing effort was to stipulate that been written about this (Hilborn et al. 1993; a certain proportion of the sardines had to be Hutchings and Myers 1994; Walters and Maguire canned. To avoid this law some manufacturers 1996; Doubleday et al. 1997; Healey 1997; Hutchings bought up old factory ships abandoned by the de- et al. 1997a,b). Conflicting evidence was available funct whaling industry and converted them into to Canadian Department of Fisheries and Oceans floating fishmeal plants. These were moored just (DFO) scientists on the state of the cod stocks at seaward of California’s three-mile state boundary, the end of the 1980s. CPUE (catch per unit of effort) allowing the producers to ignore the CFG ruling data from the commercial fishery remained high about canning a proportion of the fish. and indicated an expanded stock while the DFO’s In the late 1930s scientists at the CFG were own survey data showed the stock to be at a steady becoming increasingly concerned by the effect the level. Data from the inshore sector had shown a de- fishery was having on the stock. In the 1936–7 sea- cline since the 1970s but these data were ignored son the fishery took around 40% of the estimated (Finlayson 1994). The general opinion is that, adult stock; by the end of the fishery, in the 1944–5 at best, the institutional structure of DFO and season, it took 74% (McEvoy 1986). A Congres- of the ways in which the Canadian Atlantic sional hearing was set up in 1936 to decide whether Fisheries Scientific Advisory Committee (CAFSAC) the stock was being overfished. The two groups of worked were not able to accommodate uncertain- scientists used very similar data to argue different ty in the assessments of the cod stock. Some have things because of their uncertainty in how to inter- argued that the bureaucratic procedures of the pret events. The CFG scientists argued that there DFO suppressed the expression of uncertainty by was clear evidence that the stock was being over- DFO scientists (Hutchings et al. 1997a). At the sci- fished. The USBF scientists, who had a brief not entific level, the assessment work was done using only to conserve stocks but also to promote the the best possible methods but these are always fishing industry, argued that the evidence on going to produce results that have considerable un- overfishing was equivocal and that environmental certainty (Walters and Maguire 1996). If people at changes had brought down the sardine stock. This both the scientific and the management level had allowed the industry representatives to drive a been better able to communicate this uncertainty wedge between the scientists at the Congressional and handle it, the collapse might have been hearing, which concluded that more evidence lessened or averted. needed to be gathered. Soon thereafter, the Second Both examples show that fisheries science is not World War stimulated demand for canned sardines yet able to provide stock assessments with well- for troops. This stimulated fishing, which contin- defined statistical precision. It probably never ued until 1945 when the sardine stock collapsed will (Walters and Maguire 1996; Walters 1998). and the fishery disappeared. This means that institutional structures must be This sad story illustrates how Charles’s (1998b) created that can make decisions in the face of un- second category of uncertainty, involving the state certainty. Suggestions have been made by Hilborn of nature, prevented scientists from giving a clear and Walters (1992) and Hilborn et al. (1993). One message that the sardine stock was being over- could use a two-phase approach to assessing stocks fished. At the time there was no procedure that and deciding on which action to take. First stock CFG or USBF biologists could use to present to the assessment scientists would evaluate the possible Congressional hearing a set of possible states with states of the stock and propose how the state will possible outcomes if a range of actions could be change as a consequence of a range of actions that taken. The result was a stalemate followed by the might be possible. The evaluation of stock states loss of a resource that has never really reappeared. (the state of nature) would also include an estimate 6 Chapter 1

Table 1.1 The form proposed by Hilborn et al. (1993) for The second phase of the process would be for inspecting the possible outcomes and their values for a the fisheries managers to review and evaluate the range of policy decisions and states of nature. P policy n = output from the first phase and decide on which option n, S = state of nature m, p = probability with 1 m option to take. At this stage, imponderables that which S exists, Y = the yield expected from the m nm could not be built into the assessment and evalua- particular combination of policy and state, E(Pn) = the expected value of each policy option calculated, as tion phase would be taken into account. shown in the text.

Policy State of nature Expected value 1.7 DEVELOPMENT TO S (p ) S (p ) S (p )...S (p ) 1 1 2 2 3 3 m m INCREASE ECONOMIC EFFICIENCY OR FREEDOM? P1 Y11 Y12 Y13 ...Y1m E(P1) P2 Y21 Y22 Y23 ...Y2m E(P2) P3 Y31 Y32 Y33 ...Y3m E(P3) Much of the argument about who owns the re- ...... source and how it should be exploited derives from ...... one particular economic viewpoint. This puts effi- ...... ciency, productivity and maximization of wealth ...... P Y Y Y ...Y E(P ) as the main objectives of economic activity. Evalu- n n1 n2 n3 nm n ated in this context, fisheries are like firms and the management problem is to structure the industry so that the rental stream is maximized. Develop- of their probability. The possible outcomes that ing a fishery is seen as equivalent to making it depend upon each state would be presented in the more efficient. Hannesson’s chapter (Chapter 12, form of a table (Table 1.1) or, in more complex this volume) outlines this approach, which has led situations, a series of computer models, which to the industrialization of fisheries as fishers strive would then be presented to the decision makers. to keep costs down through economies of scale. An This group would then evaluate the costs and alternative view of development is provided by Sen benefits of each alternative and determine the risk (1999) who argues that ‘development [is] a process associated with each. The expected value of each of expanding real freedoms that people enjoy’. The possible policy in Table 1.1 is calculated from: freedom so gained allows people to live more ful- m filled lives through access to improved education, EP()nnmm= Â Y p , (1.1) health care and a clean and undamaged environ- m=1 ment. This concept does not depend on the wealth where Ynm is the expected yield from a particular of a society. Artisanal fishers on Lake Victoria, for combination of policy n and state of nature m, pm is example, have lost freedoms through the rise in the probability of the mth state of nature and E(Pn) importance of the introduced Nile perch (Lates is the expected value of choosing policy n. Part of niloticus) (Harris et al. 1995). The fishery for this the evaluation process would involve simulating species has become a capital-intensive enterprise the results of policy changes that the decision catching fish for export. Artisanal fishers, who group might come up with (‘what if’ questions). catch fish for their own consumption, have had These evaluations would take into account the their prospects diminished as the cichlid species benefit that would accrue to the fishery in all its on which they depend have become rarer (see forms. For this to be done the decision-making Reynolds et al. Chapter 15, this volume), and body would have to include representatives of all because they cannot afford to enter the Nile perch the major stakeholders and, as a group, they would fishery. This has reduced their freedom to choose have to be clear on the objectives they wanted to how to develop their lives and has deprived them of achieve. potential improvements in health, better educa- The Human Dimension of Fisheries Science 7 tion and the chance to free themselves from the from them in the way we do now or to benefit from constraints imposed by poverty. the maintenance of biodiversity. Economic efficiency and accumulation of The economic argument is set in more general wealth are often means to greater freedom but they terms. Deriving from an original analysis by are not goals in themselves. With greater freedom, Ramsey (1928), the problem is to schedule people can choose the lifestyle they want although consumption so as to maximize utility over time not without regard to the rest of society. In many modelled by the equation: senses modern economic activity, with its focus tT= on efficiency and profit, has left out the human di- -dt uuceTt= () (1.2) mension, a point eloquently put by Hoskins (1954) Ú t=0 in a discussion of the changes seen in the south

Devon, UK, fishing industry. He points out how where uT is the utility over an infinite time period small fishing communities in south Devon have T, ct is consumption at time t, u(ct) is utility as a gradually decayed as the fishing industry has be- function of consumption at time t and d is the dis- come more and more industrialized, and one of the count rate. The object is to maximize uT given the consequences has been that it is no longer possible level of consumption and the discount rate. This ‘for the small man [to] find a living and work on his approach has been used by Clark (1990) to examine own account, in his own way and at his own pace’. the effects of the discount rate on exploitation of a In other words, many people lose freedom of action fish population. He pointed out that if the interest as a result of improved economic efficiency. rate of money in the bank was greater than the rate of growth of the exploited fish population, it makes rational sense to fish the stock to extinction and invest the money earned in the bank. From the 1.8 ACHIEVING point of view of sustainable development, this SUSTAINABLE means that the present value of the future of the DEVELOPMENT fish stock is too low to preserve. Anand and Sen (1996) argue that such action still qualifies as It is now well established that the management sustainable development, so long as the money in- objectives of the early fishery scientists were too vested in the bank would give future generations narrow. It is not enough to aim for maximum the capacity to live a life comparable in opportuni- sustainable yield, and this is rarely done any more ties (freedoms) to ours in the present even though (Punt and Smith 2001). In a sense the argument has those opportunities would not be identical. Under come full circle in that the modern objective in re- this view, sustainable development would obvi- source management is sustainable development. ously not mean that the world would remain This concept was the centrepiece of the Brundt- unchanged. The freedoms would be achieved by land Commission’s report (Brundtland 1987) and converting living resources and biodiversity into the idea was accepted as a goal for most countries money. In this vein, Solow (1991) has argued that at the 1992 conference on the environment in Rio ‘[i]f you don’t eat one species of fish, you can eat an- de Janeiro, Brazil. Anand and Sen (1996) have dis- other species of fish . . .’ from which the conclu- cussed the concept in detail. The key issue is that sion is drawn that ‘we do not owe to the future any we should not do anything in the present which particular thing. There is no specific object that would deprive future generations of having the the goal of sustainability, the obligation of sustain- same opportunities as we have (the same free- ability, requires us to leave untouched’. doms, see Section 1.7). In fisheries terms this could This conclusion is predicated on the assump- mean that we should not destroy or diminish fish tion that the structure and diversity of the environ- stocks to such a degree that future generations ment are unimportant to our continued welfare. It would not have the opportunity to gain a living stems from the human-centric view of life, which 8 Chapter 1 separates human activity from the rest of nature as lifestyle opportunities as the present. Other stake- if the two had little of importance to connect them. holders would also suffer from the reduced diver- But many people feel that biodiversity has a strong sity and changed function of aquatic ecosystems. inherent value above our immediate practical In addition, as discussed by Young and Muir (Chap- needs. Furthermore, the two volumes of this book ter 3, this volume), each fish species has its own have set out to demonstrate how fisheries are em- value as a product and one cannot assume that bedded in an ecological framework and that we once all the cod have been fished, people will hap- cannot have one without the other. Chapters 11 to pily convert to eating farmed salmon. As is so often 16 in Volume 1 and Chapters 10, 14, 15 and 16 in the case, economic analysis leaves out the intangi- Volume 2 illustrate how modern knowledge forces bles that motivate people. us to recognize the critical importance of ecosys- The potential future benefits of biodiversity, tems and biodiversity to the sustainability of fish- apart from the aspect of ecosystem function, are eries. It is not true that one fish species is equally as hard to assess. It is often argued that reducing the good as another, as argued by Solow (1991). Species stock of biodiversity reduces the chances of dis- have different roles and removing one from the covering new substances that might be of benefit system can have complex and unexpected conse- in medicine or the food industry (Reaka-Kudla quences. If we damage ecosystems irreparably, et al. 1997). Because we do not know what those then the opportunities for future generations will substances might be and what the relation is be- be severely reduced. At the simplest level, people tween their existence and biodiversity, we cannot will not be able to eat cod and chips, but, more assess in an objective way what we might miss if significantly, the changed balance in the oceans, we destroy fish stocks now. We are back to the combined with continued production of CO2 into structural uncertainty that started the first chap- the atmosphere, could alter our environment so ter of the first volume. In the final analysis we radically that life as we know it could not con- probably have to accept that there will always be a tinue. Solow (1991) argues that our ancestors con- limit to our knowledge and that what we do today sumed less than they had a right to, thereby leaving will have important consequences for the present us with the means to a better life than we perhaps as well as for our legacy for future generations. deserve. We are now redressing the balance and consuming more than we should, thereby leaving our inheritors with less than they will need to live ACKNOWLEDGEMENTS a good life. We thank Tony Pitcher and Chuck Hollingworth for valuable comments on an earlier draft of this 1.9 CONCLUSIONS chapter.

It is very difficult to evaluate the importance of saving fish resources and biodiversity for the fu- REFERENCES ture. In terms of food alone, we may imagine that aquaculture could replace all the fish we presently Anand, S. and Sen, A.K. (1996) Sustainable Human capture from the wild. From a wider perspective, as Development: Concepts and Priorities. New York: we have outlined in Section 1.8, diminished popu- United Nations Development Programme. lations and damaged ecosystems will alter the Beverton, R.J.H. and Holt, S.J. (1956) The theory of fish- prospects of many aspects of our lives for the fu- ing. In: M. Graham (ed.) Sea Fisheries: Their Investiga- tion in the United Kingdom. London: Edward Arnold, ture. As it is likely that many fishers choose their Ch. IX, pp. 372–441. activity because it suits their lifestyle needs (Hart Brundtland, Gro Harlem (1987) Our Common Future: and Pitcher 1998), catching all the fish now would World Commission on Environment and Develop- not allow future generations to enjoy the same ment. Oxford: Oxford University Press. The Human Dimension of Fisheries Science 9

Charles, A.T. (1998a) Beyond the status quo: rethinking Hutchings, J.A. and Myers, R.A. (1994) What can be fishery management. In: T.J. Pitcher, P.J.B. Hart and D. learned from the collapse of a renewable resource? Pauly, Reinventing Fisheries Management. London: Atlantic cod, Gadus morhua, off Newfoundland and Chapman & Hall, Ch. 6, pp. 101–11. Laborador. Canadian Journal of Fisheries and Aquatic Charles, A.T. (1998b) Living with uncertainty in Sciences 51, 2126–46. fisheries: analytical methods, management priorities Hutchings, J.A., Walters, C.J. and Haedrich, R.L. (1997a) and the Canadian groundfish experience. Fisheries Is scientific inquiry incompatible with government in- Research 37, 37–50. formation control? Canadian Journal of Fisheries and Clark, C.W. (1990) Mathematical Bioeconomics: The Aquatic Sciences 54, 1198–210. Optimal Management of Renewable Resources. Hutchings, J.A., Haedrich, R.L. and Walters, C.J. (1997b) Chichester: Wiley. Reply: scientific inquiry and fish stock assessment in Clark, R.W. (1968) The Huxleys. London: Heinemann. the Canadian Department of Fisheries and Oceans. Doubleday, W.G., Atkinson, D.B. and Baird, J. (1997) Reply: the interplay of policy, politics, and science. Comment: scientific inquiry and fish stock assess- Canadian Journal of Fisheries and Aquatic Sciences ment in the Canadian Department of Fisheries and 54, 1430–1. Oceans. Canadian Journal of Fisheries and Aquatic Kaiser, M.J., Spence, F.E. and Hart, P.J.B. (2000) Sciences 54, 1422–6. Fishing-gear restrictions and conservation of benthic Finlayson, A.C. (1994) Fishing for Truth: A Sociological habitat complexity. Conservation Biology 14, 1512– Analysis of Northern Cod Stock Assessments from 25. 1977 to 1990. St John’s, NF: Institute of Social Lankester, E.R. (1895) The Right Hon., Thomas Henry and Economic Research, Memorial University of Huxley. Journal of the Marine Biological Association, Newfoundland. UK 4 (ns), 1–2. Gordon, H.S. (1953) An economic approach to the Lee, A.J. (1992) The Directorate of Fisheries Research: optimum utilization of fishery resources. Journal of Its Origins and Development. London: Ministry of the Fisheries Research Board of Canada 10, 442–57. Agriculture, Fisheries and Food. Graham, M. (ed.) (1956) Sea Fisheries: Their Investiga- McEvoy, A.F. (1986) The Fisherman’s Problem: Ecology tion in the United Kingdom. London: Edward Arnold. and Law in the California Fisheries, 1850–1980. Cam- Harris, C.K., Wiley, D.S. and Wilson, D.C. (1995) Socio- bridge: Cambridge University Press. economic impacts of introduced species in Lake Mangelsdorf, T. (1986) A History of Steinbeck’s Cannery Victoria. In: T.J. Pitcher and P.J.B. Hart (eds) The Im- Row. Santa Cruz: Tanager Press. pacts of Species Changes in African Lakes. London: Punt, A.E. and Smith, A.D.M. (2001) The gospel of Chapman & Hall, pp. 215–42. maximum sustainable yield in fisheries management: Hart, P.J.B. (1998) Enlarging the shadow of the future: birth, crucifixion and reincarnation. In: J.D. Reynolds, avoiding conflict and conserving fish. In: T.J. Pitcher, G.M. Mace, K.H. Redford and J.G. Robinson (eds) Con- P.J.B. Hart and D. Pauly, Reinventing Fisheries servation of Exploited Species. Cambridge: Cambridge Management. London: Chapman & Hall, Ch. 17, pp. University Press, pp. 41–66. 227–38. Ramsey, F.P. (1928) A mathematical theory of saving. The Hart, P.J.B. and Pitcher, T.J. (1998) Conflict, consent and Economic Journal 38, 543–59. cooperation: an evolutionary perspective on individ- Reaka-Kudla, M.L., Wilson, D.E. and Wilson, E.O. (eds) ual human behaviour in fisheries management. In: T.J. (1997) Biodiversity II: Understanding and protect- Pitcher, P.J.B. Hart and D. Pauly, Reinventing Fisheries ing our Biological Resources. Washington, DC: Joseph Management. London: Chapman & Hall, Ch. 16, pp. Henry Press. 215–25. Reynolds, J.D., Mace, G.M., Redford, K.H. and Robinson, Healey, M.C. (1997) Comment: the interplay of policy, J.G. (eds) (2001) Conservation of Exploited Species. politics, and science. Canadian Journal of Fisheries Cambridge: Cambridge University Press. and Aquatic Sciences 54, 1427–9. Sen, A.K. (1999) Development as Freedom. Oxford: Hilborn, R., Pikitch, E.K. and Francis, R.C. (1993) Cur- Oxford University Press. rent trends in including risk and uncertainty in stock Solow, R.M. (1991) Sustainability: an economist’s per- assessment and harvest decisions. Canadian Journal spective. Paper presented as the eighteenth J. Seward of Fisheries and Aquatic Sciences 50, 874–80. Johnson Lecture. Woods Hole Oceanographic Institu- Hilborn, R. and Walters, C.J. (1992) Quantitative tion, 14 June, Woods Hole, Mass. Reprinted in R. Fisheries Stock Assessment: Choice, Dynamics and Dorfman and S. Dorfman (eds) (1993) Economics of Uncertainty. London: Chapman & Hall. the Environment: Selected Readings. New York: W. W. Hoskins, W.G. (1954) Devon. London: Collins. Norton. 10 Chapter 1

Walters, C.J. (1998) Designing fisheries management assessment from the northern cod collapse. Reviews systems that do not depend upon accurate stock in Fish Biology and Fisheries 6, 125–38. assessment. In: T.J. Pitcher, P.J.B. Hart and D. Pauly Wilen, J.E. (1979) Fisherman behaviour and the design (eds) Reinventing Fisheries Management. London: of efficient fisheries regulation programmes. Journal Chapman & Hall, Ch. 21, pp. 279–88. of the Fisheries Research Board of Canada 36, Walters, C.J. and Maguire, J-J. (1996) Lessons for stock 855–8. Part 1 Background to Fisheries

2 Fish Capture Devices in Industrial and Artisanal Fisheries and their Influence on Management

OLE ARVE MISUND, JEPPE KOLDING AND PIERRE FRÉON

2.1 INTRODUCTION vessels, equipment for navigation and fish find- ing, of the catches taken and the costs involved Modern fishing has grown out of the methods and are so different that separate considerations are systems still used in artisanal fisheries. A 120- necessary. metre pelagic trawler operating out of Killybegs, Ireland, is different from its predecessors by virtue of the technology that has been applied to increase 2.2 MAIN FISH CAPTURE the ship’s catching power and to reduce the num- TECHNIQUES ber of people required to run it. In reality there is a 2.2.1 Introduction continuum between the most primitive and most advanced fishing equipment and this chapter will The main fishing gears have distinct constructions illustrate this development. The elaboration of and methods of operation (Fig. 2.1). Purse seines fishing gear from early times is outlined by Smith capture fish shoals by surrounding them with a (Chapter 4, this volume), and this development huge net. Trawls filter water masses at a speed through time is paralleled by the spatial changes higher than the fish’s sustainable swimming observed between fleets using a large technologi- speed. In long lining fish are attracted by the odour cal input and those using almost none. The new of baits that they swallow and get hooked. Gill- technology has clearly introduced new problems nets form invisible net walls that fish swim into to be coped with by fisheries managers and we will and get gilled or become entangled. spend some time in this chapter outlining the par- The fishing vessels from which the respective ticular management problems attached to indus- gears are operated generally differ in their con- trialized and artisanal fisheries. Along the way we struction and equipment with respect to size, en- will describe the equipment used in the various gine power and gear handling devices. Gill-netters fisheries and how it is used. haul the gear over a roller at the rail (Fig. 2.2). Be- In this chapter we focus on the main fishing hind the net hauler that is mounted to the deck, techniques operated by industrialized fleets taking fish are removed from the net by hand, and the net the bulk of the catch in oceanic fisheries, and on is cleared and stacked in a net bin manually or by the techniques employed by artisanal fishermen in automatic net clearers, ready to be set again over lakes and coastal waters in Third World countries. the stern. The principle for handling long lines is The basic principles for these methods is generally much the same – hauling over a roller at the rail – the same whether they are operated by industrial- but the fish are removed before the line goes ized fleets or artisanal fishermen, but the size around the line hauler (Fig. 2.2). The line is then of the gear, of the tools for gear handling, of the passed through a tube and back to a room with stor- 14 Chapter 2

Purse seine Trawl Gill-net

Long line

Fig. 2.1 Main fishing capture techniques. From left to right: surrounding of fish shoals by purse seine; filtration of water masses by trawl at a higher speed than the fish are able to endure; gilling of fish that move into a net; attraction and hooking of fish by baited long line.

ing magazines or onto a storing drum. The line is stern where the trawl doors hang when not in use set over the stern through a machine that auto- (Fig. 2.2). The trawl is hauled and shot through the matically baits the hooks, or else branch lines with stern ramp, and hauling and shooting are con- baited hooks are clipped on manually. Large long trolled by the sweep winches in the bow. The bot- liners have refrigerated and/or freezer holds for the tom gear is hauled into a set of trawl lanes (there bait and the catch, and also fish processing ma- are usually two such sets on modern vessels) and chines for heading, splitting and fillet production the trawl net and the bag are pulled in by aid of the (Sainsbury 1986; Bjordal and Løkkeborg 1996; gilson winches on the boat deck. A pelagic trawler Bach et al. 1999a). A purse seiner has two powerful has no gilson or sweep winches: the trawl net is winches to pull in the purse line and thereby close therefore wound up onto a powerful net drum with the net up along the side of the vessel (Fig. 2.2). The 5–50 tonnes pulling capacity. net is then hauled on board by the net winch, Modern industrialized fisheries are very effec- pulled backwards through a net tube or slide by the tive. Strict regulations through the setting of Total net crane that also stacks the net in the bin at the Allowable Catches (TACs) and technical measures stern of the vessel. The catch will finally be con- like minimum mesh size and closed areas, as well centrated along the side of the vessel in the bag of as effort limitations and control of catches and the purse seine and pumped or brailed onboard landings, are therefore needed to prevent the col- with a scoop made of net and into tanks with refrig- lapses of many economically important stocks. A erated sea water (RSW) at -1.5°C. This preserves generally important technical measure for fishing the catch at the best quality. A bottom trawler has gears is the size selectivity, which is defined as the two powerful winches for towing and hauling the proportion of fish retained related to the length of trawl gear with 5–50 tonnes pulling capacity, and the fish (Anon. 1996). A selection curve for trawl the trawl warp passes through towing blocks at the gears is mostly sigmoid (see Sparre and Hart, Fish Capture Devices 15

Gill-netter 24 × 7 m Engine Fish 500hp casing hold Cost: $US 2 million Net Compartments bin Gill-net Working deck hauler

Long liner 34 × 8 m 800hp Cost: $US 5 million Working deck Compartments Freezer Bait Fish Line hatch hold hauler Line magazines Store Setting Line machine Purse seiner 55 × 11 m 3000hp Cost: $US 10 million Net Fish holds (refrigerated sea water) stacking crane Compartments Purse winches Net bin Net slide Fish pump Net winch Ring needle Bottom trawler 60 × 13 m 5000hp Cost: $US 20 million Trawl winch Deck store Compartments Sweep

Stern Fish hatch Trawl ramp lanes

winches Trawl winch Compartments Deck compartments

Fig. 2.2 Principal deck layout of main types of fishing vessels (provisional vessel size, engine power and building costs are indicated).

Chapter 13, this volume, equation 13.11), but bell- retained, and L25 is fish length where 25% of the shaped curves can be the case for gill-nets and fish is retained),these parametres describe the size hooking gears. Important selectivity measures are selection characteristics of fishing gears.

L50, defined as the fish length where 50% of the fish is retained by the gear, and the selection factor de- 2.2.2 Purse seining fined as L50 divided by mesh size in centimetres. In addition to the selection range which is defined as The principle of purse seining is to surround fish

L75 - L25 (L75 is fish length where 75% of the fish is shoals by a large net that can be closed so that the 16 Chapter 2

fish cannot escape (Ben Yami 1990). When the net higher than in shoals. Normally, purse seining on is hauled back, the fish will become concentrated shoals takes place in darkness during night-time, in a bag ready for brailing or pumping onboard. The while fishing on schools is limited to the daylight first purse seines were developed by Rhode Island hours. In some fisheries, the fish are available to fishermen in the USA for catching menhaden profitable purse seining both when schooling (Brevoortia patronus) in the 1860s. during daytime and when shoaling at night. For The purse seine is kept floating by a line of floats example, this is usually the case during the winter at the surface, and the lower part of the net sinks by fishery for capelin (Mallotus villosus) off the coast the force of a heavy leadline. The net will thus be of northern Norway, and on the spawning grounds stretched out as a circular wall surrounding the of Norwegian spring-spawning herring off the fish shoal (Fig. 2.1). The mesh size is so small that coast of western Norway in winter. Other purse the net wall acts as an impenetrable fence prevent- seine fisheries are profitable only when the fish is ing escape. When the purse seine has been set out shoaling at night or when schooling during day- and allowed to sink for some minutes, so that it time. An example of the former is the once-large reaches deeper than the depth of the target fish Chilean fishery for Chilean jack mackerel (Trachu- shoals, it can be closed by hauling the purse line. rus murphyi), which normally is conducted when The lower part of the purse seine will be confined the fish occur in dense shoals near the surface at and pulled to the surface. When this operation is night (Hancock et al. 1995). Most of the purse seine finished it is impossible for fish shoals to escape, as fisheries for herring and mackerel in the North Sea long as the net is not torn. However, flying fishes in summertime are conducted when the fish are (Exocoetidae) or mullets (Mugilidae) can still jump schooling during the daylight hours. over the floatline. The fishing capacity of purse seiners is normal- The size of the purse seine depends upon the be- ly proportional to vessel size. In the Chilean jack haviour of the fish to be captured and the size of the mackerel fishery where there were no limitations vessel from which it is operated. For catching fast- set by fishing quotas in the mid-1990s, the total swimming fish shoals at depth, purse seines must annual catch of purse seiners was related to the be long, deep, have a high hanging ratio and be hold capacity of the vessel through the equation: heavily leaded. The hanging ratio is defined as the total annual catch (tonnes) = 33.3¥hold capacity length of the stretched net divided by the length (m3) + 18.2 (Hancock et al. 1995). In the 1992 of the line on which it is mounted. For catching season, a purse seiner with a hold capacity of slower-swimming fish shoals distributed near the 1350m3 was able to land about 65000 tonnes of surface, purse seines can be shorter, shallower, and Chilean jack mackerel! have a low hanging ratio. The relationships be- In some regions, artificial light is used to attract tween fish species, vessel size and purse seine char- fish at night. When sufficient fish have aggregated acteristics are given in Fig. 2.3. near the light source, they are caught by purse sein- Modern purse seining is mostly dependent on ing (Ben-Yami 1971). This technique is probably detection and location of fish shoals by hydro- of greatest importance for purse seine fisheries acoustic instruments (Misund 1997). Larger in Asian countries, where it is used offshore. The purse seiners (>40m) have a low-frequency, low- technique is also common in the Mediterranean, resolution sonar (18–34kHz) for detecting fish the Black Sea, and in the Russian and African shoals at long range, and a high-frequency, high- lakes. In other regions, the technique is mostly resolution sonar (120–180kHz) for more detailed used inshore, as during the sprat (sprattus sprat- mapping of shoal size and fish behaviour in rela- tus), herring (Clupea harengus) and saithe (Pol- tion to the vessel and the net. lachius virens) fisheries in the fjords of southern Purse seining is conducted on fish aggregated in Norway. dense shoals (Pitcher 1983), or on fish occurring in Tuna purse seining is conducted by large vessels distinct schools, in which the density is much (mostly > 60m) with large nets (Fig. 2.3) in Fish Capture Devices 17

Sprat 450 m ρ = 0.45 22 mm 90 m 17 x+ 5m 3.0 kg m–1

Anchoveta 600 m ρ = 0.20 15 mm 80 m 45 x+ 10m 5.0 kg m–1

Capelin, Herring, and Mackerel 650 m ρ = 0.45 22–35 mm 160 m 6.0–7.0 kg m–1

60 x+ 12m Jack mackerel 1000 m

ρ = 0.40 35 mm 200 m

7.0 kg m–1

Tuna 2000 m

ρ = 0.25 150 mm 200 m

1.5 kg m–1

70 x+ 13m

Fig. 2.3 Relationships between fish species, vessel size and purse seine characteristics (length, depth, hanging ratio (r), mesh size in mm and lead weight on ground rope are indicated).

tropical/subtropical regions in the Atlantic, Indian Kaiser and Jennings, Chapter 16, this volume). Ocean, particularly in the Mozambique Channel, Usually, large tuna are present underneath the and around the Seychelles, off Australia, and in the dolpins. When the purse seine is closed around the western and eastern Pacific (Fonteneau 1997). In encircled dolpins and tuna, the fishers attempt to the three oceans, the tuna is caught by sets made release the dolphin by the backdown procedure. on free-swimming schools or by sets made on fish This causes the floatline of the distant part of the associated with floating objects of natural or artifi- purse seine to submerge so that the dolphins can cial origin, which are mostly trees and branches. In swim and jump over. However, dolphin frequently the eastern Pacific, tuna are also caught by sets become entangled in the purse seine and drown. made on dolphin herds with which the tuna is In total, this amounted to about 60000 killed associated (Anon. 1992; Hall 1998). The dolphin dolphins due to the tuna purse seine fishery in the herds are visible on the surface, and rather easy to eastern Pacific in 1992 (Anon. 1992). The tuna encircle by the fast-going tuna seiners (see also pursers operating in this region have been under 18 Chapter 2 pressure to change the fishing strategy or fishing such as grenadier (Macrouridae) in the slopes of operation to eliminate the accidental killing of the deep-water basin of the Atlantic and hoki dolphins (Kaiser and Jennings, Chapter 16, this (Macruronus novaezelandiae) and orange roughy volume). in New Zealand and Australian waters. The ratio of warp length-to-bottom depth decreases from 2.2.3 Trawling about 10:1 in shallow waters to 3:1 at intermedi- ate depths, and further down to 1.8:1 when towing In principle, trawling is filtration of water masses at great depths. inhabited by fish at a higher speed than the target The construction of bottom trawls reflects the fish species are able to swim sustainably. The type of species targeted and the bottom type to be filtration is done by towing a net bag horizontally operated on. Trawls used on sandy bottoms or soft through the water masses, either in the pelagic sediments, such as industrial trawling in the zone (pelagic trawling) or along the bottom (bot- North Sea or shrimp trawling in Arctic waters, tom trawling). often have no rollers on the ground gear. On such trawls the ground gear is just a wire wound with Bottom trawling rope. On the other hand, trawls to be operated on A bottom trawl is usually rather heavy, with two hard bottoms with stones, sponges and hard corals steel doors to open the trawl and with enough are equipped with heavy ground gear with steel or weight to keep the trawl in contact with the bot- rubber bobbins. During the last decades heavy tom during towing (Fig. 2.1). The doors, which can rock-hopper gear with rubber discs have become be up to 2m high and 4m long and weigh up to 7 popular in the fleets of white-fish trawlers operat- tonnes, are connected to the vessel by the trawl ing in North Atlantic waters. With this ground warp, usually a wire up to 32mm in diameter on gear trawlers are able to fish on hard bottom sub- the largest trawlers. The trawl bag is connected to strates with obstacles such as stones and rocks the doors by a set of one to three bridles on each without damaging the gear or suffering snags on side, and in the mouth of the trawl there is a ground bottom obstacles. gear with steel or rubber bobbins or rollers that In the southern North Sea beam trawls are com- enable the gear to be operated on rough and stony monly used for catching flatfishes such as sole grounds without being torn or hooked. The first (Solea solea) and plaice (Pleuronectes platessa). proper bottom trawls, with wooden otter boards These trawls are kept open by an iron boom up to and rollers on the ground line, were constructed about 10m wide that runs on iron shoes on each in Scotland in the 1890s (Smith, Chapter 4, this side so that the boom is kept about 60cm above the volume). A long time before that, however, fisher- bottom during towing. The beamers can operate men on sailing ships in the North Sea and in the one trawl on each side of the vessel. A beam con- English Channel had been towing a trawl bag on nected to the bow mast, which can be turned at 90° soft or sandy bottom that was kept open by a to the side of the vessel, is used to handle the boom wooden boom. and the attached trawl net. The beamers are usual- Bottom trawling is conducted in depths as shal- ly powerful vessels with up to 4500hp available, low as about 20m when trawling for industrial and can be up to 50m in length. They can tow the species such as the sand lance (Ammodytes sp.) gear at a speed up to 6 knots. in the North Sea, at intermediate depths of about With an otter trawl, fish entering between the 250m when trawling for cod and haddock trawl doors tend to be guided by the bridles to- (Melanogrammus aeglefinus) in the Barents Sea, in wards the mouth of the trawl, where they turn and deep waters from 400m to 800m for Arctic try to swim in front of the rolling ground gear. shrimps (Pandalus borealis) and orange roughy Small fish and species with poor swimming capac- (Hoplostethus atlanticus) off Namibia, and in ity like plaice are less herded than larger and faster depths from 1000 to 2000m for deep-water species fish. As fish get exhausted they turn back and enter Fish Capture Devices 19 into the trawl bag, but in the net section leading In recent years considerable focus has been to the bag or in the bag itself, the optomotoric given to the physical impact of bottom trawl gears response causes the fish to turn forward and try on the bottom fauna and topography (Kaiser and to swim along with the moving mesh wall. Fish Jennings, Chapter 16, this volume). On sandy bot- finally panic when in the bag and swim towards, toms, tracks of heavy beam trawls were shown to and try to escape through, the meshes. The selec- have faded completely after 37 hours (Fonteyne tivity of the trawl bag then occurs as small fish 2000). On other bottom types with more fragile pass through the meshes while a greater propor- fauna the impact can be more permanent. In the tion of the larger fish are physically unable to do so. southern North Sea several benthic species have In most bottom trawl fisheries there are regula- decreased in abundance and even disappeared from tions on minimum mesh size in the bag of the certain regions (Bergman and van Santbrink 2000). trawl, also called the cod-end, to enable proper se- Some regions of deep-water coral reefs (Lophelia lectivity of target species. Trawl selectivity is also sp.) off western Norway have been ‘clear cut’ with affected by the number of meshes in the circumfer- heavy rock-hopper trawl gears (Fosså et al. 2000). ence of the trawl bag, the length of the extension, the length of selvage ropes relative to trawl bag and Pelagic trawling extension, twine thickness, catch size, and mesh geometry. In the North Sea whitefish fishery trawl Pelagic species are also caught in large quantities selectivity can be enhanced by changing the geom- by pelagic trawling, both with single boats and by etry of the meshes in the cod-end from rhombic to pair trawling. In the Atlantic, capelin (Mallotus square (Robertson and Stewart 1988; van Marlen villosus), herring, horse mackerel, mackerel, sar- 2000), while in other areas use of such a mesh con- dines and sprat are caught by pelagic trawling. Off figuration did not improve selectivity because of Ireland there is a large pelagic trawl fishery for clogging by redfish (Sebastes spp.) (Isaksen and blue whiting (Micromesistius poutassou), which Valdemarsen 1989) or catch size-dependent selec- aggregate for spawning during winter and spring. tivity (Suuronen et al. 1991). Use of rigid sorting In the northern Pacific, there is a large pelagic grids, made of steel or aluminium, inserted into trawl fishery for Alaska pollock (Theragra the net, gives a sharper and more efficient selection challcogramma). than mesh selection alone (Larsen and Isaksen Single-boat pelagic trawling was mainly devel- 1993). Fishers in the Barents Sea whitefish trawl oped after the Second World War, and is one of the fishery have been required by regulation to use most sophisticated fishing techniques, which sets grids since 1997. specific demands for the size, power, equipment In the relatively fine-meshed shrimp trawls, by- and operation of the vessels. In the Netherlands, catch has been a substantial problem. There are es- there has been the development of a fleet of super- timates of bycatch to shrimp catch ratios of 5:1 in trawlers of up to about 125m in length, and with a temperate waters and up to about 20:1 in tropical carrying capacity of up to 7000 tonnes of frozen waters (Pender et al. 1992; Alverson and Hughes pelagic fish. These vessels are operated both in the 1996; Ye et al. 2000). To reduce bycatch in shrimp northern and southern Atlantic, fishing a variety trawls, the Nordmore grid was made mandatory in of species such as horse mackerel, mackerel, her- the Barents Sea in 1990 (Isaksen et al. 1992). The ring and sardine. During the last decade a substan- name ‘Nordmore’ derives from Nordmøre, a tial fleet of pelagic trawlers, which are often also northern district in Norway where the idea for the rigged for purse seining, has been built in Northern sorting grid originated. Other bycatch reduction Europe (Ireland, Norway, Scotland). These vessels devices (mesh or grid constructions) have been de- are 60–80m long, with main engines from 5000 to veloped for tropical shrimp trawling, such as in the 12000hp, and a carrying capacity of 1500–2500 northern prawn fishery off Australia (Broadhurst tonnes of fish in refrigerated seawater (RSW) et al. 1997; Brewer et al. 1998; Salini et al. 2000). tanks. 20 Chapter 2

In principle, a pelagic trawl is a net bag, towed The mesh size of pelagic trawls can be up to tens of behind a single vessel or between two vessels oper- metres in the front part, but decreases gradually to ating together. During pair trawling, the two boats a few centimetres in the bag. The meshes in the both pull the trawl, and open it horizontally by front part of the trawl must herd the target fish into going parallel to each other, but some distance the centre of the trawl opening. If these meshes are apart (500–1000m). A single-boat trawl is kept too large, the target fish can escape, and the catch- open by the lateral forces of two large trawl doors ing efficiency of the trawl decreases. The meshes (5–15m2) in front of the trawl. The doors are at- in the cod-end must be so small that it is physically tached to the warps from the vessel, and the trawl impossible for the target fish to escape. is connected to the doors via a pair of two or more Pelagic trawling is conducted mainly on fish sweeps. The length of the sweeps depends on the occurring in large shoals, extended aggregations vertical opening of the trawl, and is about 180m for and layers. The fish can be recorded by sonar or a 30m high trawl. A pair of weights (50–2000kg) at- acoustically, and the trawl opening is normally tached to the lower wings and the weight of the monitored by a cable-connected net sonde or a doors pull the trawl downwards. The fishing depth trawl sonar. These instruments provide informa- of the trawl is adjusted by the warp length, the tow- tion on the opening of the trawl, and the presence ing speed and the vertical inclination of the doors. of fish inside or outside the trawl opening. There At a towing speed of about 6ms-1 (3.5 knots), a are also acoustic sensors to monitor the door warp length of about 500m gives a fishing depth of spread, the headline height and depth, and catch about 200m for a trawl with 30m vertical opening, sensors that are activated when there is catch in 1000kg weights on the lower wings, and with the the cod-end. doors weighting about 3300kg (Valdemarsen and Misund 1995). In most cases there are also floats or 2.2.4 Long lining kites attached to the headline to give the trawl an upward pull. On single boats, using pelagic trawls A longline is made up of a mainline to which designed to catch small pelagic fish such as branch lines with baited hooks are attached at capelin, which are about 15cm in length, there can regular intervals (Fig. 2.1). The mainline can be an- be a pair of extra doors attached to the upper wings chored to the bottom, and is sustained by vertical to give the trawl a proper opening. Such trawls are float lines equipped with instrumented buoys towed at a low speed of about 3ms-1, and the later- with light, radar reflector and radio beacon. The al pull of just two doors will be too little to give the total extended length of the gear varies from a few trawl the intended opening. According to the size, hundred metres to more than 150km and is made the vertical opening of a pelagic trawl varies by of several subunits called skates or baskets about an order of magnitude, from about 15 to (Yamaguchi 1989a; Bjordal and Løkkeborg 1996). 150m. The horizontal opening is usually about Three main types of longline can be identified (see equal to the vertical, and the total area of the Table 2.1), according to their position in the water opening of pelagic trawls thus varies by about two column, the type of filament used in the mainline, orders of magnitude, from about 200 to 20000m2. hook spacing and the shape of the hook, which can Pelagic trawls are constructed according to spe- be J-, wide gape, circle or EZ baiter (which has an cific combinations of mesh size, twine, tapering intermediate shape between the circle hook and and panel depth. Normally, the trawls are con- the J-hook and the name implies the ease with structed of two or four panels that are joined in the which the bait can be attached): selvage or laced together. The size of the trawl is • bottom longline, which is subdivided into given as the circumference of the trawl opening. demersal and semipelagic; This is calculated as the number of meshes in the • pelagic or drifting longline. trawl opening, minus the number of meshes in the The rigging of longlines varies according to the tar- selvage, multiplied by the stretched mesh size. get species, fishing conditions and countries. A de- Fish Capture Devices 21

Table 2.1 Main characteristics of the different types of depth in Antarctic waters (Bjordal and Løkkeborg longlines. 1996). Pelagic long lining targets tunas world wide Gear category Mainline Hook Hook type (Fonteneau 1997). Bigeye (Thunnus obesus) and spacing (m) yellowfin (T. albacares) tuna are mainly caught Bottom along the whole intertropical areas of all oceans Demersal Multifilament 1.2 to 1.8 EZ baiter plus in the northern part (from 20° to 40°N) of the Semipelagic Monofilament 2 to 3 J- or wide gap Pacific Ocean for bigeye only. Albacore (T. alalun- Pelagic Multifilament 30 to 100 J- or circle ga) is primarily caught in the northern (25° to 35° then hook N) and southern (10° to 40°S) part of the oceans monofilament while bluefin tuna (T. thynnus) are mainly ex- ploited in the southern hemisphere (20° to 50°) on each side of the African continent and around Australia and New Zealand (Yamaguchi 1989c; mersal longline is usually made of a multifilament Fonteneau 1997; Carocci and Majkowski 1998). mainline of 4–11mm in diameter, more resistant Subsurface longlines target swordfish (Xiphias to chafing than the monofilament lines. The line gladius) in warm waters. lays on the sea bed with an anchor. The branches, which are also called snoods, are made either from Bait and baiting multifilament or monofilament, and are some- times ended with steel wire or chain to prevent A variety of bait types are used by demersal long cutting by predators such as sharks. Semipelagic lining, such as small pelagic fish, saithe, shrimp, longlines are also anchored but the mainline is crab, octopus, mussels, whelks and lugworms. In maintained above the bottom by alternate floats pelagic long lining only the thawed saury (Colo- and sinkers set at regular intervals. Modern pelagic labis saira) was used until recent years, but the bait longlines are usually made of a synthetic, mono- tends nowadays to be more diversified (sardine, filament mainline of 3–4mm in diameter. For tuna squid, etc.). The bait is usually dead, whole or cut fishing, monofilament branch lines of about 20m in pieces. Artificial bait are currently being tested length and 2mm diameter are attached to but are not commonly used at present. Chemical the mainline by metallic snaps. The rigging of the light sticks are sometimes added to the gear in the branches can be made more sophisticated by swordfish fishery. Baiting is done manually before the use of swivels, aimed to limit twisting and setting, either onshore or on board, or by a baiting chafing, and a combination of monofilament, machine during the setting process. Automated multifilament and coated steel wires (Cook 1989; baiting can be subdivided into precise baiting, Yamaguchi 1989a,b; Bach et al. 1999a). where a piece of bait fish is automatically cut and hooked, and random baiting where the longline passes through a container with a mixture of pre- Catch distribution cut bait and water. In tuna long lining, prebaited Demersal long lining occurs mainly in the north branch lines are stored in tubs or boxes and are at- Atlantic, targeting cod (Gadus morhua), haddock, tached to the mainline during setting. The use of tusk (Brosme brosme) and hake (Merluccius sp.) baiting machines is common on board bottom long and in the northeastern Pacific where it targets liners larger than 30m as these require the baiting essentially cod (G. macrocephalus) and halibut of up to several tens of thousands of hooks. Manual (Hippoglossus stenolepis). The common depth of baiting is common in coastal and artisanal exploited ground varies from 100 to 800m, but can fisheries. be deeper as in the case of the Patagonian toothfish (Dissostichus eleginoides), exploited up to 2500m 22 Chapter 2

than multifilament ones, and the attraction of Setting and hauling some species, such as swordfish, to light sticks Setting the longline is normally done from the (Boggs 1992; Bjordal and Løkkeborg 1996; Bach et stern of the vessel. The time of setting depends on al. 1999b). the target species but also on the feeding habits of The success of long lining largely depends on bait scavengers. The end marker buoy and floats fish behavioural aspects: level of hunger, diel for- are dropped first at low speed, followed by the buoy aging habits, prey attack, abundance and patchi- line and the anchor in the case of bottom longlines. ness of the natural prey in the habitat. In French Then the vessel speeds up for setting the rest of the Polynesia, for instance, bigeye tuna catch rate is, as gear. Pre-baited bottom lines can be set at speeds of expected, positively correlated to the large-scale up to 10 knots since only the different skates need abundance of the micronekton. Nevertheless to be linked. When snoods are snapped on the line, within relatively rich areas this correlation is a timer secures an even hook spacing. Three to five negative at the much smaller scale of the long- men are employed during the shooting. line where the micronekton distribution is patchy. Hauling is usually performed from the side of This is interpreted by a competition between the vessel, starting by the marked buoy and the natural prey and baits (Bertrand 1999). anchor for bottom longlines. When intermediate buoys and sinkers are used, they must be detached Species and size selectivity from the mainline. In tuna long lining, snoods are also detached when they arrive on board. More than 30 pelagic species can be caught by pelagic longlines and the question of untargeted species mortality by hooking is a major concern, The importance of fish behaviour mostly for shark species. Selectivity of longlines The capture process of long lining is based on the depends first on the match between the fish habi- feeding behaviour of the fish (see Juanes et al., tat and the place where the gear operates, both in Chapter 12, Volume 1). The bait is regarded by the the horizontal and vertical plane. Second, the type fish as a food item, detected at long distance by the of bait used favours some species or sizes over odour trail that is dispersed by the water current. others due to preference for food items (Juanes et al., The chemical senses of olfaction and taste are there- Chapter 12, Volume 1). Finally, the size and type of fore essential and allow detection from several hook affects the species composition, because of hundred metres. Vision can only detect objects differences in feeding behaviour. The bait size, and from a few metres. Fish are able to locate the bait to a lesser extent the hook size, are positively cor- by swimming upstream. The current intensity and related with the fish size (Bjordal and Løkkeborg direction play a major role in the shape and exten- 1996), as expected from typical predator–prey size sion of the so-called odour plume released by the correlations (Juanes et al., Chapter 12, Volume 1). bait. The rate of release of attractants, which are However, bottom longlines seem to catch about amino acids, in the water decreases rapidly with the same size range of cod as bottom trawls with time so that strength has been reduced to one half 135mm mesh size, and smaller fish than those after one hour, and therefore fishing time usually caught by gill-nets (Huse et al. 2000). For Green- ranges from 10 to a maximum of 24 hours and can land halibut (Reinhardtius hippoglossoides), long- be reduced to only 2–3 hours when hooked fish can lines catch a size range larger than that caught by be attacked by large predators. Vision is an impor- bottom trawls, but smaller than by gill-nets (Huse tant sense at short range, especially in midwater, et al. 1999). and explains why some species attack the bait mainly during the phase of deployment or hauling Impact on sea birds of the gear. Vision also explains the higher catching rate of monofilament lines, which are less visible Impacts of fisheries on sea birds are reviewed by Fish Capture Devices 23

Kaiser and Jenings (Chapter 16, this volume). We source of food (Bjordal and Løkkeborg 1996; therefore do not go into the details here, except to Nambiar and Sudari Pawiro 1998). In Japan, some reiterate the scale of the problem and methods of tuna species like bluefin and bigeye are tradi- reducing mortality. Up to 70% of the baits can be tionally eaten raw (‘sashimi’ and ‘sushi’) and taken by birds during longline setting. Estimated regarded as a delicacy. The wholesale price paid at values of the number of birds killed per 1000 hooks landing for top sashimi-quality tuna can reach range from 0.36 to 0.44 depending on both longline more than US$200 per kilogram. However, only area and bird species, which results in the killing of very small quantities sell at these high prices and several thousands of birds per year (Ryan and Boix- tuna used for canning can be sold for less than Hinzen 1998). Moloney et al. (1994) used an age- US$1 per kilogram (Carocci and Majkowski 1998). structured model to simulate population trends of Long lining is a cost-effective way of catching albatross (Diomedea exulans). The simulation re- fish because it does not require a lot of energy, nor a sults portray a population decreasing at a rate of large crew. This is especially so in coastal fishing 2.3% per year, presumed to be a result of deaths that does not involve long travel distances. Never- caused by longline fishing vessels on this long- theless the gear is relatively expensive, as is the lived species. bait and the baiting process, which is either labour- To limit bird mortality, different methods are intensive when manual or needs costly invest- used such as shooting lines during the night, which ment in baiting machines. The advantages of mainly reduces the capture of albatross but is less longlines over other fishing gears are: the quality of effective for petrel. It is also possible to fit lines the fish which are often alive when caught; no or with lead weights, aimed to obtain a faster sinking, little impact on the sea bed; little ‘ghost fishing’ by or to scare the bird away from the line. The two lost equipment; little fish dumping; a relatively most efficient methods consist in either using good size selectivity, limiting juvenile catches and a setting tunnel or trailing a ‘bird scarer’ made of the possibility to reduce unexpected mortality a line with pendants parallel to the longline by using an adapted fishing strategy (Bjordal and (Løkkeborg 1998). Løkkeborg 1996; Bach et al. 1999c).

Commercial aspects 2.2.5 Gill-netting Longlines mostly target highly priced species like Hand-knitted nets of natural fibres like cotton are tuna and demersal species (Table 2.2), which are among the oldest and most widespread fishing very important economically and a significant gears. Gill-netting is still amongst the most impor-

Table 2.2 Characteristics of the main longline fisheries.

Gear Common range Main target species Yearly catches Main fishing category of boat size (m) during the 1990s countries (million tonnes)

Bottom 8–55 Cod, hake, haddock, tusk, 0.5 Norway, halibut, Patagonian toothfish Iceland, Russia, Spain, Argentina Pelagic 25–60 Bigeye, yellowfin and bluefin 0.4 Japan, tuna, albacore Taiwan, South Korea 24 Chapter 2 tant fishing methods, and is used to harvest a vari- actively moving around, either searching for food ety of species worldwide. The nets are now made of or migrating, and they function best at night when synthetic fibres like polyamide forming monofila- the fish do not see the webbing. Multifilament ment or multifilament twines of which the net is nets are less efficient than monofilament nets knitted. There are several types of twine: twisted (Washington 1973; Hylen and Jackobsen 1979). multifilament, monofilament, monotwine and Experiments have shown that attaching baits to multimono. The webbing of gill-nets should be gill-nets may increase catches of gadoids such as as transparent and invisible as possible because tusk and ling (Molva molva) (Engås et al. 2000). In gill-nets are passive gears, the fish itself moving areas with a high density of gill-nets, such as into the net webbing head on, and trying to push during spawning aggregations of cod in Lofoten, through the mesh opening. Small fish pass Norway, in winter, a satiation effect can occur so through, but the larger ones with a maximum cir- that there will be a decreasing catch rate with an cumference just bigger than the girth of the mesh increasing number of nets (Angelsen and Olsen opening become gilled. The very large fish can 1987). Currents can reduce net height and change entangle themselves or escape capture. the shape of the webbing and thereby affect the To keep it standing as a rectangular net wall in efficiency substantially (Stewart and Ferro 1985). the sea, the net is mounted under a floatline with A model of how various factors influence catch positive buoyancy and a groundline with negative rates of bottom-set gill-nets for cod has been devel- buoyancy. The floatline can be made of a floating oped (Dickson 1989a). material like polypropylene, and may have syn- Fish caught in gill-nets can be gilled, wedged or thetic floats embedded in it or solid floats formed entangled. In hard and stiff monofilament nets cod as rings or rectangular pieces can be mounted are mainly gilled, while in soft multifilament nets on it. The groundline can have solid lead pieces cod are mainly entangled (Stewart 1987). Despite mounted externally, or small lead pieces can be this, gill-nets are very size-selective; few fish are embedded in the groundline itself. For pelagic caught differing from optimum length by more drifting gill-nets the positive buoyancy of the float- than 20% (Hamley 1975). The optimum girth for line is larger than the negative buoyancy of the capture is about 1.25 times the mesh perimeter and groundline, while the opposite is the case for the optimum length for capture increase approxi- bottom-set gill-nets. At each end the gill-net webbing mately in proportion to mesh size (Engås and is mounted to a breast line joining the float and Løkkeborg 1994). A loosely hung net catches fish groundline. Gill-nets are normally joined together over a much wider size range than a more tightly to form fleets that can consist of 10–100 units. Im- hung net. portant parametres determining the size of fish to In recent years there has been much focus be captured are the mesh size and the hanging ratio on ‘ghost fishing’ by lost gill-nets. (Kaiser and of the gill-net webbing. Gill-nets are usually rather Jennings, Chapter 16, this volume). This is because selective, but bycatch of sea birds and marine synthetic gill-nets can continue to fish for several mammals can be substantial in some fisheries. years after the nets have been lost. In Norway there Trammel nets, which have two large-meshed and a is an official programme to dredge for lost gill-nets small-meshed net webbing in between have low in areas where fishermen report loss of nets or selectivity. This is because such nets capture fish where it is known that there is much gill-netting. by entangling. Also, bycatch of sea mammals such as dolphins, The visual stimulus of the net is a key factor de- harbour porpoise or seals take place in gill-nets termining whether or not fish detect the net (Cui (Tregenza 2000; Kaiser and Jennings, Chapter 16, et al. 1991), and at a certain light threshold gill-nets this volume). become invisible, depending on colour, net ma- terial and turbidity (Dickson 1989b). Gill-nets are therefore normally set to operate when the fish are Fish Capture Devices 25

2.3 ARTISANAL FISHERIES fisheries are their numerous actors (Fig. 2.4), re- sulting in an extraordinary variety in terms of spe- There is no standard, or even clear definition of cialized capture solutions to different resources, small-scale artisanal fisheries (Panayotou 1982; environments and seasons, with many being only Ben Yami 1989; Bâcle and Cecil 1989; Durand et al. seasonal in combination with agriculture, still 1991) other than perhaps ‘scale’ and ‘artisanal’, small-scale fisheries are highly dynamic and where the latter means a craftsman relying on diverse in terms of fishing techniques within each his skill by himself, or with the help of family exploited ecosystem (e.g. Gerlotto and Stéquert members or a few companions. Generally, arti- 1978). sanal fisheries can be seen as an integrated infor- Apart from the purely technical classification mal way of living within a geographically limited of capture devices (see von Brandt 1959, 1984; community and intrinsically dependent on local Nédélec 1975; Nédélec and Prado 1990), small- resources, rather than a formal occupation with scale fishing gear can be grouped into appliance or a broad spectrum of options in terms of fishing development categories (Ben-Yami 1989): grounds, markets and alternative investment 1 Traditional, ‘primitive’ fishing implements and opportunities (Thomson 1980; Panayotou 1982; installations characteristic for artisanal subsis- Bâcle and Cecil 1989). tence fisheries in developing countries. These are locally produced and operated from small canoes 2.3.1 Diversity of methods and rafts, and by wading or diving fishers. Barriers, weirs, traps, pots, spears, tongs, harpoons, gill- The modern large-scale fishing techniques that nets, dip-nets, small beach seines, and simple have been described so far have all developed from hook-and-line gear are included in this group. small-scale equivalents, and given the evolution in 2 Modern, ‘sophisticated’ equipment used almost power, mobility and technical solutions of fishing exclusively in industrialized countries. For vessels, the congruent evolution in commercial example dredges, automated haulers and jiggers, capture techniques have mainly been in terms of hydroacoustic and electronic equipment, and size, refinement and power requirements, with various auxiliary labour-saving devices. an ensuing increasing dependency on modern 3 Intermediate equipment used by small-scale port facilities. Thus, small-scale artisanal capture fishermen in both industrialized and developing techniques comprise all the categories and com- countries. This group consists of most fishing nets binations of fishing methods already described such as gill-nets, seines, and lift nets and much of (von Brandt 1959, 1984; Nédélec 1975; Nédélec the hook-and-line gear, almost all of which are fac- and Prado 1990) but are different by being less tory produced and made of synthetic materials. capital-, vessel-, and fuel-intensive than modern Modern accessories, however, such as mechanized industrialized fishing methods (Fig. 2.4). Many do net and line haulers, light lures, and echo sounders not require vessels or energy-consuming acces- are increasingly used in developing countries, and sories at all but can be operated from the shore, and particularly by the small-scale enterprises which many have remained virtually unchanged for cen- expand into harvesting previously underutilized turies. Others are in rapid technological develop- pelagic species in lakes, reservoirs and coastal ment, like the Senegalese pelagic canoe fisheries fisheries. (WWF 1998), and may only be distinguished as ‘ar- tisanal’ by the scale of the fishing units, and their 2.3.2 Research problems in still limited range. As a consequence a further dis- artisanal fisheries tinction between commercial- and subsistence- artisanal fisheries has therefore sometimes been One of the first striking problems in artisanal fish- used (Kesteven 1976; Smith 1979). An important eries is the lack of data and scientific literature. characteristic, however, of small-scale artisanal About 25–30% of the world total output of fish, or 26 Chapter 2

Large-scale Small-scale company-owned artisanal

Number of fishermen employed Over 12 000 000 Around 450 000

Marine fish caught for human consumption Around 24 million tonnes annually Around 20 million tonnes annually

$$$$$$$$$ Capital cost of each $$$$$$$$ job on fishing vessel $$$$$$$$ $ $100 to $1000 $10 000 to $100 000

Bycatch discarded at sea Around 20 million tonnes annually Around 1 million tonnes annually

Marine fish caught for industrial reduction to meal and oil, etc. Around 19 million tonnes annually Almost none

Fuel oil consumption 10 to 14 million tonnes annually 1 to 2 million tonnes annually

Fish landed per tonne = = of fuel consumed 2 to 5 tonnes 10 to 20 tonnes

Fishermen employed for each $1 million invested in fishing vessels 10 to 100 1000 to 10000

Fig. 2.4 Comparison of large-scale commercial fisheries with small-scale artisanal fisheries. (Source: modified from Thomson 1980.)

nearly half of the landings for consumption Platteau 1989). Considering the magnitude of this (Panayotou 1982; Fitzpatrick 1989; Allsopp 1989), sector, the relative dearth of literature compared come from small-scale fisheries, which engage with industrial fisheries is astounding and severe. 80–90% of all fishermen, estimated at between 12 Moreover, most literature on artisanal fisheries and 15 million people (Smith 1979; Alsopp 1989; is (semi)-anthropological or socioeconomic and Fish Capture Devices 27 gives, if anything, only descriptive accounts of cap- often up to 20m length, is used to carry the catch ture devices and species compositions (e.g. Durand (Fréon and Weber 1983). Similar artisanal fisheries et al. 1991). The lack of quantitative data is a occur along the coast of West Africa (Bard and particularly acute problem for making meaningful Koranteg 1995), particularly in Ghana and Côte suggestions for research and management (Larkin d’Ivoire. The Ghanaian canoe fleet has over 8000 1982), but this may also be one of the reasons for units, of which more than 2000 are purse-seining, the many unsolicited notions that exist around but the yield is lower than in Senegal (around artisanal fisheries. It may be fair to state that 70000 tonnes per year) due to lower fish density. small-scale fisheries have received only scant In Venezuela the pelagic artisanal fishery is also attention during the past few decades of national well developed and targets mainly S. aurita using and international development, with the excep- encircling nets operated by up to five canoes. In tion of some agencies specialized in research in contrast with the conventional purse seine, which developed countries, such as the Institute of was commonly used until recently, this net does Research for Development (IRD, formerly not close at the bottom. The fish school is encir- ORSTOM), the International Center for Living cled and then the net is pulled to the shore where Aquatic Resources Management (ICLARM), and its bottom part is in contact with the sea floor. The Food and Agriculture Organization (FAO). net is then secured by anchors and buoys, and left Many aquatic ecosystems are exploited by for up to a week for a larger vessel to extract the artisanal fisheries only. Coastal lagoons, tidal flats, fish still alive and transport them to canneries and shallow shores, estuaries and coral reefs on the fishmeal plants. The landings commonly reach marine side, as well as most freshwater fisheries 140000 tonnes per year in recent years (Fréon and such as lakes, reservoirs, rivers and floodplains, are Mendoza, in press). by their size, depth, topography or inaccessibility The common limitation of artisanal fisheries is only suited for small-scale enterprises and are the range of operation. In Senegal two new ways of exploited in a wide variety of ways. Other coastal exploiting resources located far from the landing areas, except for uninhabited coastlines such as sites have developed from the 1970s. The first one, those found off Namibia, are mostly exploited by called ‘marée pirgoque’, consists in using large both artisanal fisheries inshore and industrial fish- canoes loaded with ice and food that travel up eries offshore. In many of these there is persistent to 800km from their landing site for a period of competition and conflicts between the two types 10 to 30 days. The second consists of loading up and, with few exceptions, the artisanal fisheries to 45 open canoes on board a large vessel that are struggling to survive against the pressures of transports them to the fishing grounds where they modernized capture techniques and overexploita- are released for line fishing over several days. The tion (Crean and Symes 1996). mother vessel provides them with food and water In Senegal, the artisanal pelagic fishery has been and stores the catches. This mode of exploitation landing over 200000 tonnes of fish per year since resembles the ‘Dory’ fishery for cod in the north- the beginning of the 1990s and competes seriously west Atlantic during of the nineteenth and early with the industrial fishery (Fréon et al. 1978; twentieth centuries (Charles-Dominique and Samb and Samb 1995). The bulk of the catch is Mbaye 1999). made up of sardines (Sardinella aurita and S. These fisheries are all based on traditional maderensis) caught by small purse seines which wooden canoes that have been modified by the are typically 250 ¥ 30m and surrounding gill- addition of an outboard engine, a synthetic hull, nets of 200m ¥ 9m. (Sech 1980); 14–16m length storage and conservation of the fish, and for the use wooden canoes, motorized with outboard engines, of modern gear. This shows the high potential for transport these gears. All the fishing operations are adaptation and evolution of this traditional fishing performed manually by a crew of 12 to 18 people. In vessel. Nevertheless the design of the canoes the case of the purse seine fishery, a second canoe, themselves, which dates back many hundreds of 28 Chapter 2 years in response to local constraints imposed by natural tropical lakes. The reported inland fish the sea and by the coastline where they are built production south of the Sahel was about 1.3 and stored, tends to remain the same and often million tonnes in 1987 (Vanden Bossche and fishermen resist changes in design. In most of Bernacsek 1990), which constituted a bit more these fisheries, a large part of the catch is processed than 40% of the total world production. Still, the to fishmeal, the price of which is largely dependent reported inland production may only be about half on the global market of soya and fishmeal (Durand the estimated potential. The three largest lakes, 1995). Therefore some crises observed in the Victoria, Tanganyika and Malawi, alone cover a artisanal fishery can find their origin on the global combined area of 132500km2 and give an annual market (Fréon and Weber 1983). yield of more than 300000 tonnes of fish (Kolding For coral reef fisheries the main gears used are 1994). The traditional artisanal inshore fisheries of nets, hook and line, and spearing or traps for finfish these and other lakes or reservoirs have over the (Medley et al. 1993). Nets are usually set just off the past three decades mainly expanded and evolved reef and fish are driven to them. Hooks and spear- towards the productive offshore resources, con- ing are more effective for larger teleost predators. sisting of small barbus and clupeid species with a Traps are the least selective, taking a variety of fish very high biological overturn. With the subse- and crustaceans depending on their construction. quent technological development of a highly spe- An important form of exploitation is gathering, cialized night fishery using larger vessels, light often combined with diving, for the more seden- attraction and various encircling seine techniques tary organisms. Unfortunately, a number of reef- or large winch-operated dip-nets, some of these destroying fishing methods are in increasing use fisheries are now rapidly crossing the border worldwide, often stimulated by the aquarium between subsistence fishing and small-scale trade (Rubec 1986). Many poisons, such as bleach commercial, or even semi-industrial, enterprises. or sodium cyanide, as well as explosives, used Rivers and floodplain fisheries are among the mainly on schooling species, not only kill the most diverse in the world and may still be catego- target fish, but also the corals (McManus et al. rized as belonging mainly within the subsistence 1997). Muro-ami is another destructive method sector of artisanal fisheries. In Bangladesh, where boulders are dropped on the corals to scare Indonesia and Thailand, for example, each river and drive fish out of the reef crevices and into may commonly be fished by 20 or more different nets. Such methods are giving rise to much inter- gears (MRAG 1994; Hoggarth et al. 1999a,b). In national concern and inevitably create a negative the Mekong Delta, producing around 1.2 million impression of the whole artisanal sector. tonnes annually, at least 62 different fishing tech- Freshwater fisheries, which account for about niques have been classified (Claridge et al. 1997), 10% of the world total capture, are predominantly under 19 general headings ranging from scoop-, artisanal and these are as diverse as the freshwater cast-, lift- and gill-nets, over a variety of traps, habitats. Gill-nets, seines, traps, hooks and lines baskets, enclosures, fences and weirs, to spears, are generally used in most lakes and reservoirs all guns, poison and explosives. A similar diversity is over the world for inshore and demersal species. found in Africa and Latin America. This general However, in most freshwater fisheries of the picture of a wide variety in fishing gears enables Western world the traditional methods are slowly the capture of the many different fish species and phasing out in favour of recreational fisheries for sizes, in the many different habitats and during the economic and social-political reasons. Important various changing seasons. As a result floodplain artisanal freshwater fisheries are therefore mainly fisheries are in general highly productive, where found in the tropical and subtropical hemisphere, yields may reach more than 100kgha-1 yr-1 with- which produce at least 90% of the inland fisheries’ out signs of biological overfishing (Kolding et al. yield. Africa is the only continent with large 1996; Hoggarth et al. 1999a). Fish Capture Devices 29

2.3.3 Myths and misconceptions gies are more socially and economically rational related to artisanal fisheries and efficient (Thomson 1980; Fréon and Weber 1983; Dioury 1985; Durand et al. 1991) and that Due to the increased marginalization of small- they may reconcile high returns on capital, low scale fisheries in the industrial world, or their investments, labour-intensiveness and high added conversion into semi-industrial or recreational value (Fig. 2.4). Finally, they constitute a prerequi- activities, artisanal fisheries are more and more as- site for the new paradigm for community-based sociated with developing countries, and are often resource management (Verdeaux 1980). Many are perceived as a traditional or even antiquated, not declining but advancing, renewing equipment, poorly equipped subsidence activity (Dioury 1985; transforming their boats and gear, and taking Bâcle and Cecil 1989; Platteau 1989). By their in- advantage of market changes (Durand et al. 1991). formal, free and liberal position they are habitually Still, the overwhelming opinion among many looked upon as unruly members of a society, actors in the fisheries sector, particularly among which are difficult to manage. Moreover, many of assessment biologists, economists and policy the attempts to ‘modernize’ these fisheries have makers, is that small-scale fisheries in developing failed due to the very fact that the desired changes countries employ inefficient, wasteful and indis- from ‘traditional’ practices have encountered criminate fishing practices (Chou et al. 1991), formidable obstacles because such changes also ignore gear regulations and legislation (Panayotou affect the social structures which are entrenched 1982; Gulland 1982), are subject to open-access in a community (Allsopp 1989; Durand et al. 1991). ‘Malthusian overfishing’ (Pauly 1994; Amar et al. Africa, for example, is ‘littered’ with scrambled 1996), and therefore need to be managed and trawlers and abandoned ice plants/cold stores controlled one way or the other. Are these notions (Bataille Benguigui 1989; Bernascek 1989) as a true, or are they myths and misconceptions? As result of well-meant, but poorly planned develop- this chapter is dealing with capture devices, we ment programmes and the lack of clearly defined will discuss here only the rationality of prohibiting government policies and commitments. However, non-selective fishing methods which is occurring improving the standard of living of small-scale in most artisanal fisheries. fishermen is but one of the objectives in fisheries policy. Competing, and often more important, ob- 2.3.4 From gear to management jectives are employment creation, and increased production for consumption, export and, not least, Capture devices are intrinsically associated with national revenue (Smith 1979; Panayotou 1982; selectivity, and selectivity, or the impact of fishing Dioury 1985; Chauveau and Samba 1989; Platteau on an ecosystem, is an essential component of a 1989). Consequently, as other economic sectors management programme (Pauly and Christensen, are ‘advancing’, many small-scale fisheries are Chapter 10, this volume). Consequently, much left behind with an increasingly negative image effort is devoted to investigations of the efficiency of being persistently poor, ignorant, ancestral, and and selectivity of active and passive fishing gears often resource-depleting or ecosystem-destructive (Fitzpatrick 1989). Recent developments have (Panayotou 1982; Bâcle and Cecil 1989; Amar et al. been motivated and promoted by world and 1996), although the latter may be a direct conse- market opinion, leading to devices that allow quence of the unfair competition they are exposed turtles to escape from shrimp trawls,that reduce to (Thomson 1980). This picture, however, is not entanglement of mammals in drift-nets and purse uniform and there is an increasing attention to the seines, and that decrease the large volumes of by- importance of this sector and a growing awareness catch in the shrimp- and other specialized fisheries of the need to conserve these communities. (see the purse seine and trawling sections above Several studies show that smaller-scale technolo- and Kaiser and Jennings, Chapter 16, this volume). 30 Chapter 2

The importance of selectivity is therefore rooted principle ecosystem conserving. All species are in most researchers and managers, and any non- preyed upon at various rates during their lifespan, selective capture method automatically carries the and for teleosts the highest mortality is usually connotation of being harmful, bad or destructive, during the early life history phase (Myers, Chapter or will at least lead to growth-overfishing, seen 6, Hutchings, Chapter 7, both Volume 1). Thus from the traditional single-species perspective theoretically, the ‘utopian’ but optimal exploitation (Shepherd and Pope, Chapter 7, this volume). pattern, by which an ecosystem could be main- Mesh size- and gear restrictions are among the tained in balance, is fishing each trophic level and most easily applied and widely used management fish population in proportion to the rate of natural regulations. Consequently most nations have mortality (Mi) it is subjected to (Caddy and Sharp imposed legislation, which bans certain gears and 1986; Kolding 1994). However, as all fishing gears mesh sizes, with the aim of protecting the resource are more or less species- and/or size-selective, such (Gulland 1982). Although many of these regula- non-selective ‘utopian’ exploitation patterns can tions have originated from the problems asso- be achieved only by employing a multitude of ciated with the large-scale fisheries, they are often gears simultaneously. The multigear, multi- uniformly applied on all sectors. However, selec- species artisanal floodplain fisheries often seem tivity seems much more a problem for indus- to be producing an overall species, abundance, and trialized fisheries, which on average dump about size composition that closely matches the ambient 45% of their catch, while small-scale artisanal ecosystem structure (MRAG 1994; Kolding et al. fisheries discard on average only 5% (Bernacsek 1996; Claridge et al. 1997; Chanda 1998; Hoggarth 1989), despite the fact that they mostly operate et al. 1999a,b). On the ecosystem level such an ex- in more multispecies environments. Although ploitation pattern may be considered unselective numerous authors have already pointed to the across the species diversity range, and floodplain problems of defining the ‘right’ mesh size in a fisheries, particularly in Asia, seem to have per- multispecies fishery, the notion of regulations on sisted, with the only change coming from natural selectivity still persists. In addition, small-scale fluctuations, despite a very high fishing effort. fisheries often use a variety of gears, both traditional Judgement on this is limited by the length of the and intermediately modern. Many of these gears, data series we have. and particularly the traditional ones such as seines, In other words, the multi-gear, which is gen- small mesh sizes, drive- or beat fishing, barriers, erally unselective, fishing pattern employed in weirs, are often classified as illegal under the pre- many small-scale fisheries, combined with the text of being non-selective with assumed negative ability of fishermen to change their target species impacts on the fish populations. However, the ac- within a single trip (Laloë and Samba 1990), is the tual impact of these methods is rarely investigated closest example of the optimal exploitation and the true aim of the regulation may be to protect pattern that exists. Therefore, the established fish- for political or social reasons the position in a ing practices versus legal frameworks may easily fishery of another less efficient gear (Panayotou become not only an infinite tug of war, as seen in 1982). In the few instances where the actual impact most instances where small-scale fisheries have of non-selective illegal gear used in small-scale resisted the implementation of gear-restrictive fisheries has been studied, it is in reality an open regulations, but also a completely futile tug of question how ‘detrimental’ these fishing methods war, as seen from the perspective of ecosystem are. Two such case studies are from the floodplain conservation. fisheries in Zambia (Kolding et al. 1996; Chanda 1998), and the artisanal beach seine fishery on the 2.3.5 Can we manage Cape coast in South Africa (Bennett 1993; Clark small-scale fisheries? et al. 1994a,b; Lamberth et al. 1995a,b,c). In fact, non-selective harvesting patterns are in The survival of traditional management sys- Fish Capture Devices 31 tems, or perhaps more commonly, the persistence and livelihoods, is used to brace and support of small-scale fisheries under unmanaged condi- the views on this group as being backwards, indis- tions, is challenged and often overwhelmed by the criminate and destructive in their fishing patterns. tendencies of modernization, accumulation and Nevertheless this sector proves that it is still concentration in which increased capital invest- capable of reacting and innovating at the level of its ment and the creation of global markets affect the organization. means of production. Accelerating technological sophistication in capture methods, as well as sub- sequent increasing conflicts between commercial 2.4 CONCLUSIONS and small-scale fisheries, may be seen as the capitalistic way of delaying the economic conse- The gear used in industrial fisheries is charac- quences of declining resources (Crean and Symes terized by changes brought about by technological 1996), as evidenced by the growth of advanced innovation. This, coupled with the intense gear technology and distant water fishing fleets competition between fishers to catch their share (Alverson and Larkin 1994). The universal solu- of the catch, has led to larger boats, nets and elec- tion to these conflicts seems to be characterized by tronic fish-finding equipment. The emphasis in the intervention of the state in terms of techno- the past has been on improving the catching power cratic regulation systems based on mesh and gear and efficiency of vessels and fishing gear. The in- restrictions, TACs and quotas (Beverton 1994). creasing demands of management are now leading However, when both artisanal and industrial fish- to the use of technology to improve the selectivity ermen are exploiting the same resource, a total of fishing, with devices such as the sorting grid, quota is certain to favour the large-scale fisheries, inserted into the trawl tunnel, helping to reduce which have greater catching capacity. Unavoid- bycatch. Further devices, such as satellites, will ably, the livelihood of the small-scale fishery will be increasingly used to monitor fishing effort be reduced while no economic surplus would be spatially and temporarily, so making it easier to generated in the fishery as a whole (Panayotou ensure that fishers are following the regulations, 1982). Unfortunately, science is used (and shaped?) There is an urgent need to remedy the lack of to legitimate the implementation of these regula- quantitative information on small-scale fisheries. tion systems and thus shield the policy makers In terms of capture devices there is a particularly from confronting the basic sociological objectives important need to establish the selectivity of the of fisheries management. By focusing on single- various gears individually and in combination in species TACs, quotas, and enhanced selective order to evaluate their impact on the ecosystem. harvesting strategies and gear development, the Faced with the complexity and heterogeneity of mainstream research is better suited for those the operating conditions of small-scale fisheries, having the equipment and vessel technology to the present national and international efforts in make a quick kill on targeted species, than for the small-scale fisheries seem concentrated on man- small-scale fishermen working outside the global agement implementations. However, without the markets. Incidentally, the dominating scientific necessary fundamental research, these implemen- paradigms thus protect and favour the interests of tations will be based on the reiterated dogma and the corporate structures and do little to investigate may face the same failures as the many previous the truth behind the possible myths and miscon- attempts to ‘modernize’ these fisheries. ceptions related to small-scale artisanal fisheries, as evidenced by the relative absence of scientific literature on this sector. On the contrary, the ACKNOWLEDGEMENTS independent, and often technically ‘illegal’ re- sponse, by small-scale fisheries to increased The authors are grateful to Pascal Bach for his valu- regulations in order to safeguard their traditions able help in the preparation of the longline section, 32 Chapter 2 and to Bente Karin Hoddevik and Anne Britt Skaar Bard, F.-X and Koranteg, K.A. (eds) (1995) Dynamics and Tysseland for preparation of the figures. Use of Sardinella Resources from Upwelling off Ghana and Ivoiry Coast. Paris: Séminaires ORSTOM. Bataille Benguigui, M.-C. (1989) La pêche artisanale aux îles Tonga: antagonisme entre projets de développe- REFERENCES ment et traditions. Aquatic Living Resources 2, 31–43. Bennett, B.A. (1993) The fishery for white steenbras Anon. (1992) Dolphins and the Tuna Industry. Litognathus lithognatus off the Cape coast, South Washington, DC: National Academy Press. Africa, with some considerations for its management. Anon. (1996) Manual of Methods of Measuring the Selec- South African Journal of Marine Science 13, 1–14. tivity Power Fishing Gears. ICES Coop. Res. Report Ben-Yami, M. (1971) Fishing with Light. Oxford: Fishing No. 215. News Books. Allsopp, W.H.L. (1989) Sustained social benefits from Ben-Yami, M. (1989) The role of small-scale fishing gear diversification of small-scale fisheries. In: Proceedings and techniques in development: challenges towards on the World Symposium on Fishing Gear and Fishing the year 2000. In: Proceedings on the World Sympo- Vessel Design. St John’s, New Foundland: Marine sium on Fishing Gear and Fishing Vessel Design. St Institute, pp. 216–26. John’s, New Foundland: Marine Institute, pp. 449–53. Alverson, D.L. and Hughes, S.E. (1996) Byctach: from Ben-Yami, M. (1990) Purse Seining Manual: FAO Fishing emotion to effective natural resource management. Manual. Oxford: Fishing News Book. Reviews in Fish Biology and Fisheries 6, 443–62. Bergman, M.J.N. and van Santbrink, J.W. (2000) Fishing Alverson, D.L. and Larkin, P.A. (1994) Fisheries science mortality of populations of megafauna in sandy sedi- and management in the 21st century. In: C.W. ments. In: M.J. Kaiser and S.J. de Groot (eds) Effects of Voigtlander (ed.) The State of the World’s Fisheries Fishing on Non-Target Species and Habitats. Oxford: Resources: Proceedings of the World Fisheries Congress, Blackwell Science, pp. 49–68. Plenary Sessions. Oxford: IBH Publishing, pp. 150–67. Bernacsek, G.M. (1989) Improving fisheries development Angelsen, K. and Olsen, S. (1987) Impact of fish density projects in Africa. In: Proceedings on the World and effort level on catching efficiency of fishing gear. Symposium on Fishing Gear and Fishing Vessel Fisheries Research 5, 271–8. Design. St John’s, New Foundland: Marine Institute, Amar, E.C., Cheong, R.M.T. and Cheong, M.V.T. (1996) pp. 155–64. Small scale fisheries of coral reefs and the need for Bertrand, A. (1999) Le système thon – environnement community-based resource management in Malison en Polynésie Française: caractérisation de l’habitat Island, Philippines. Fisheries Research 25 (3–4), pélagique, étude de la distribution et de la capturabilité 265–77. des thons par méthodes acoustiques et halieutiques. Bach, P., Wendling, B., Misselis, C. and Abbes, R. (1999a) Thesis, Université de Bretagne Occidentale, France. Forme et comportement de la palangre dérivante Beverton, R.J.H. (1994) The state of fisheries science. monofilament. In: R. Abbes and F.-X. Bard (eds) Etude In: C.W. Voigtlander (ed.) The State of the World’s du Comportement des Thonidés par l’Acoustique et la Fisheries Resources: Proceedings of the World Pêche en Polynésie Française: Rapport final Conven- Fisheries Congress, Plenary Sessions. Oxford: IBH tion Territoire/EVAAM/IFREMER/ORSTOM, no. Publishing, pp. 25–54. 951070, pp. 289–360. Bjordal, Å. and Løkkeborg, S. (1996) Longlining. Oxford: Bach, P., Misselis, C. and Abbes, R. (1999b) Les interac- Fishing News Books. tions entre la palangre et les ressources thonières Boggs, C.H. (1992) Depth, capture time, and hooked hauturières. In: R. Abbes and F.-X. Bard (eds) Etude du longevity of longline-caught pelagic fish: timing bites Comportement des Thonidés par l’Acoustique et la of fish with chips. Fisheries Bulletin (US) 90 (4), Pêche en Polynésie Française: Rapport final Con- 642–58. vention Territoire/EVAAM/IFREMER/ORSTOM, no. von Brandt, A. (1959) Classification of fishing gear. In: H. 951070, pp. 394–435. Kristjonsson (ed.) Modern Fishing Gear of the World. Bach, P., Dagorn, L. and Misselis, C. (1999c) Species selec- Oxford: Fishing News Books, pp. 274–96. tivity factors in longline fisheries: results from experi- von Brandt, A. (1984) Fish Catching Methods of the ments in the Society archipelago. SCTB 12, WP/RG-2. World, 3rd edn. Oxford: Fishing News Books. Bâcle, J. and Cecil, R. (1989) La pêche artisanale en Brewer, D.T., Rawlinson, Eayrs, S. and Burridge, C. (1998) Afrique Sub-Saharienne: Sondages et rescherches. An assessment of bycatch reduction devices in a tropi- Hull: Agence Canadienne de Dévelopment Interna- cal Australian prawn trawl fishery. Fisheries Research tional (ACDI). 36, 195–215. Fish Capture Devices 33

Broadhurst, M.K., Kenelly, S., Watson, J.W. and Clark, B.M., Bennett, B.A. and Lamberth, S.J. (1994b) Workman, I.K. (1997) Evaluations of the Nordmore Assessment of the impact of commercial beach-seine grid and secondary bycatch-reducing devices (BRDs) netting on juvenile teleost populations in the surf in the Hunter River prawn-trawl fishery. Australian zone of False bay, South Africa. South African Journal Fisheries Bulletin 95, 210–19. of Marine Science 14, 255–62. Caddy, J.F. and Sharp, G.D. (1986) An ecological frame- Cook J. (1989) The Hawaii tuna longline fishery: a 10- work for marine fishery investigations. FAO Fisheries year history, 1979–1989. SPC/Fisheries/WP16. Technical Paper. No 283. Rome: FAO. Crean, K. and Symes, D. (eds) (1996) Fisheries Manage- Carocci, F. and Majkowski, J. (1998) Atlas of Tuna and ment in Crisis. Oxford: Fishing News Books. Billfish Catches. Rome: FAO, CD-ROM. Cui, G., Wardle, C.S., Glass, C.W., Johnstone, A.D.F. and Chanda, B. (1998) Effects of weir fishing on commercial Mojsiewicz, W.R. (1991) Light level thresholds for fish stocks of the Bangweulu swamp fisheries, Luapula visual reaction of mackerel, Scombrus scombrus L., province, Northern Zambia. M.Phil thesis, Depart- to coloured monofilament nylon gillnet materials. ment of Fisheries and Marine Biology, University of Fisheries Research 10, 255–63. Bergen. Dickson, W. (1989a) Cod gillnet simulation model. Charles-Dominique, E. and Mbaye, A. (1999) Les usages Fisheries Research 7, 149–74. de l’espace dans la pêche artisanale sénégalaise. Com- Dickson, W. (1989b) Cod gillnet effectiveness related munication au 4ème forum halieumétrique, Rennes, 29 to local abundance, availability and fish movement. June to 1 July 1999. Fisheries Research 7, 127–48. Chauveau, J.P. (1991) The historical geography of Dioury, F. 1985. Insights on the development aspect and fisheries migrations in the CECAF region (end 19th future importance of artisanal fisheries. ICLARM century to 1980s). Fishermen’s Migrations in West Newsletter July 1985, vol. 8 (3), 16–17. Africa, FAO/DANIDA/NORWAY, IDAF/WP/36. Durand, M.-H. (1995) Mode of price formation for pelagic Rome: FAO, pp. 12–35. species and exploitation prospects in the less produc- Chauveau, J.P. and Samba, A. (1989) Market Develop- tive upwelling zone. In: F.-X Bard and K.A. Koranteg ment, Government Interventions and the Dynamics (eds) Dynamics and Use of Sardinella Resources from of the Small-Scale Fishing Sector: An Historical Upwelling off Ghana and Ivoiry Coast. Paris: Perspective of the Senegalese Case, Development ORSTOM Editions, pp. 194–204. and Change, London, Vol. 20, October, pp. 599– Durand, J.R., Lemoalle, J. and Weber, J. (1991) La 629. recherche face à la pêche artisanale (Research and Chou, L.M., Chua, T.E., Khoo, H.W., Lim, P.E., Paw, J.N., small-scale fisheries). Symposium International Silvestre, G.T., Valencia, M.J., White, A.T. and Wong, ORSTOM-IFREMER, Montpellier, France, 3–7 July P.K. (eds) (1991) Towards an integrated management 1989. Paris: Orstom Editions, collection Colloques et of tropical coastal resources. ICLARM Conference Séminaires. Proceedings No. 22. National University of Singapore, Engås, A. and Løkkeborg, S. (1994) Abundance estima- National Science and Technology Board, Singapore, tion using bottom gillnet and longline – the role of fish and ICLARM, Manila. behaviour. In: A. Fernø and S. Olsen (eds) Marine Fish Claridge, G., Sorangkhoun, T. and Baird, I. (1997) Com- Behaviour in Capture and Abundance Estimation. munity Fisheries in Lao PDR: A Survey of Techniques Oxford: Fishing News Books, pp. 134–65. and Issues. IUCN, The World Conservation Union, Engås, A., Jørgensen, T. and Angelsen, K.K. (2000) Effects Vientiane, Lao PDR. on catch rates of baiting gill-nets. Fisheries Research Clark, B.M. and Bennett, B.A. (1993) Are juvenile fish an (in press). issue in the trek net controversy? In: L.E. Beckley Fitzpatrick, J. (1989) Fishing technology, fisheries re- and R.P. van der Elst (eds) Fish, Fishers and Fisheries: sources and future demands. In: Proceedings on the Proceedings of the Second South African Marine World Symposium on Fishing Gear and Fishing Vessel Linefish Symposium, Durban, 23–4 October 1992. Design. St John’s, New Foundland: Marine Institute, Special Publication, Oceanographical Research pp. 1–11. Institute, Durban. Fonteyne, R. (2000) Physical impact of beam trawls on Clark, B.M., Bennett, B.A. and Lamberth, S.J. (1994a) A seabed sediments. In: M.J. Kaiser and S.J. de Groot (eds) comparison of the ichthyofauna of two estuaries Effects of Fishing on Non-target Species and Habitats. and their adjacent surf zones, with assessment of the Oxford: Blackwell Science, pp. 15–36. effects of beach seining on the nursery functions of Fonteneau, A. (1997) Atlas des pêcheries thonières tropi- estuaries for fish. South African Journal of Marine cales, captures mondiales et environnement. Paris: Science 14, 121–31. Orstom Editions. 34 Chapter 2

Fosså, J.H., Morkensen, P.B. and Furevik, D.M. (2000) Asian floodplain river fisheries. Part 2: Summary of Lophelia coral reefs in Norway: distribution and ef- DFID research. FAO Fisheries Technical Paper 384/2. fects of fishing. Firken og havel, 2. Rome: FAO. Fréon, P. and Mendoza, J. (eds) (in press) La sardina Huse, I., Gundersen, A.C. and Nedreaas, K. (1999) Rela- (Sardinella aurita), su medio ambiente y su ex- tive selectivity of Greenland halibut (Reinhardtius plotación en el oriente de Venezuela. Brussels: hippoglossoides, Walbaum) by trawls, longlines and European Union. gillnets. Fisheries Research 44, 75–93. Fréon, P., Stéquert, B. and Boely, T. (1978) La Pêche des Huse, I., Løkkeborg, S. and Soldal, A.V.S. (2000) Effects of poissons pélagiques cotiers en Afrique de l’Ouest des selectivity and fishing strategy in trawl, longline and îles Bissagos au nord de la Mauritanie: description des gillnet fisheries for cod and haddock. ICES Journal of types d’exploitation. Paris: ORSTOM Editions, série Marine Science 57, 1271–82. Océanographie, 18 (16), 209–28. Hylen, A. and Jackobsen, T. (1979) A fishing experiment Fréon, P. and Weber, J. (1983) Djifère au Sénégal: la pêche with multifilament, monofilament and monotwine artisanale en mutation dans un contexte industriel. gill nets in Lofoten during the spawning season of Revue des Travaux de L’Institut Des Pêches Maritimes Arcto-Norwegian cod in 1974. Fiskeridirektoratets 47 (3–4), 261–304. Skrifter Serie Havundersøkelser 16, 531–50. Garcia, S. and Reveret, J.P. (1991) Recherche er structure Isaksen, B. and Valdemarsen, J.W. (1989) Selectivity des pêches artisanales: Paradigmes et méthodes de experiments with square mesh codends in bottom recherche. Une introduction. In: J.R. Durand, J. trawls, 1985–1987. In: Proceedings of the Square Mesh Lemoalle and J. Weber (eds) La recherche face à la pêche Workshop held at the World Symposium on Fishing artisanale. Symposium international ORSTOM- Gear and Fishing Vessel Design, St John’s, Canada, IFREMER, Montpellier, France, 3–7 July 1989. Paris: November 1988, ed. H.A. Carr. Boston, MA: Marine ORSTOM Colloques et Séminaires. Fisheries Institute. Gerlotto, F. and Stéquert, B. (1978) La Pêche maritime Isaksen, B., Valdemarsen, J.W., Larsen, R. and Karlsen, L. artisanale en Afrique de l’ouest: caractéristiques (1992) Reduction of fish bycatch in shrimp trawl using générales. La Pêche Maritime, May 1978. a separator grid in the aft belly. Fisheries Research 13, Grainger, R. (1999) Recent Trends in Global Fishery 335–52. Production. www.fao.org/fi/trend/catch/catch.asp. Kesteven, G.L. (1976) Resources availability related to Gulland, J.A. (1982) The management of tropical multi- artisanal fisheries. In: T.S. Estes (ed.) Proceedings of the species fisheries. In: D. Pauly and G.I. Murphy (eds) Seminar-Workshop on Artisan Fisheries Development Theory and management of tropical fisheries. and Aquaculture in Central America and Panama. ICLARM Conference Proceedings No. 9. Manila: International Center for Marine Resources Develop- ICLARM, pp. 287–98. ment. Kingston, RI: University of Rhode Island, pp. Hall, M. (1998) An ecological view of the tuna-dolphin 130–42. problem: impacts and trade-offs. Reviews in Fish Kolding, J. (1994) Plus ça change, plus c’est la même Biology and Fisheries 8 (1), 1–34. chose. On the ecology and exploitation of fish in fluc- Hamley, J.M. (1975) Review of gillnet selectivity. Journal tuating tropical freshwater systems. DSc. thesis, of the Fisheries Research Board of Canada 31, Department of Fisheries and Marine Biology, Universi- 1943–69. ty of Bergen. Hancock, J., Hart, P.J.B. and Antezana, T. (1995) Kolding, J., Ticheler, H. and Chanda, B. (1996) Assess- Searching behaviour and catch of horse mackerel ment of the Bangweulu Swamps fisheries. Final Report (Trachurus murphyi) by industrial purse-seiners off prepared for WWF Bangweulu Wetlands Project, south-central Chile. ICES Journal of Marine Science SNV/Netherlands Development Organization, and 52, 991–1004. Department of Fisheries, Zambia. Hoggarth, D.D., Cowan, V.J., Halls, A.S., Aeron-Thomas, Kurlansky, M. (1999) Cod: A Biography of the Fish that M., McGregor, J.A., Garaway, C.A., Payne, A.I. and Changed the World. London: Vintage. Welcomme, R.L. (1999a) Management guidelines Laloë, F. and Samba A. (1990) La pêche artisanale au for Asian floodplain river fisheries. Part 1: A spatial, Sénégal: ressource et stratégies de pêche. Paris: hierachical and integrated strategy for adaptive co- ORSTOM, Etudes et Thèses. management. FAO Fisheries Technical Paper 384/1. Lamberth, S.J., Bennett, B.A. and Clark, B.M. (1995a) The Rome: FAO. vulnerability of fish to capture by commercial beach- Hoggarth, D.D., Cowan, V.J., Halls, A.S., Aeron-Thomas, seine nets in False Bay, South Africa. South African M., McGregor, J.A., Garaway, C.A., Payne, A.I. and Journal of Marine Science 15, 25–31. Welcomme, R.L. (1999b) Management guidelines for Lamberth, S.J., Clark, B.M. and Bennett, B.A. (1995b) Fish Capture Devices 35

Seasonality of beach-seine catches in False bay, Pauly, D. (1994) Small-scale fisheries in the tropics: South Africa, and implications for management. South marginality, marginalization, and some implications African Journal of Marine Science 15, 157–67. for fisheries management. In: E.K. Pikitch, D.D. Lamberth, S.J., Bennett, B.A., Clark, B.M. and Janssens, Huppert and M.P. Sissenwine (eds) Global Trends: P.M. (1995c) The impact of beach-seine netting on the Fisheries Management. American Fisheries Society benthic flora and fauna of False Bay, South Africa. Symposium, vol. 20, pp. 40–9. South African Journal of Marine Science 15, 115–22. Pender, P.J., Willing, R.S. and Cann, B. (1992) NPR Larkin, P.A. (1982) Directions for future research in bycatch a valuable resource? Australian Fisheries 51 tropical multispecies fisheries. In: D. Pauly and G.I. (2), 30–1. Murphy (eds) Theory and Management of Tropical Pitcher, T.J. (1983) Heuristic definitions of shoaling Fisheries. ICLARM Conference Proceedings, no. 9. behaviour. Animal Behaviour 31, 611–13. Manila: ICLARM, pp. 309–28. Platteau, J.P. (1989) The dynamics of fisheries develop- Larsen, R. and Isaksen, B. (1993) Size selectivity of rigid ment in developing countries: a general overview. sorting grids in bottom trawls for Atlantic cod (Gadus Development and Change 20, 565–97. morhua) and haddock (Melanogrammus aeglefinus). Robertson, J.H.B. and Stewart, P.A.M. (1988) A com- ICES Marine Science Symposia 196, 178–82. parison of the size selection of haddock and whiting Løkkeborg, S. (1998) Seabird by-catch and bait less in by square and diamond mesh codend. Journal Conseil long-lining using different selling methods. ICES international Exploration Mer 44, 148–61. Journal of Marine Science 55, 145–9. Rubec, P.J. (1986) The effects of sodium cyanide on coral MacLennan, D.N. and Simmonds, E.J. (1992) Fisheries reefs and marine fish in the Philippines. In: J.L. Acoustics. London: Chapman & Hall. Maclean, L.B. Dizon and L.V. Hosillos (eds) Proceed- McManus, J.W., Reyes, R.B. and Nanola, C.L. (1997) ings of the First Asian Fisheries Forum. Manila: Asian Effects of some destructive fishing methods on coral Fisheries Society, pp. 297–302. cover and potential rates of recovery. Environmental Ryan, P.G. and Boix-Hinzen, C. (1998) Tuna longline Management 21 (1), 69–78. fisheries off southern Africa: the need to limit seabird Medley, P.A., Gaudian, G. and Wells, S. (1993) Coral reef bycatch. South African Journal of Fisheries Science fisheries stock assessment. Reviews in Fish Biology 94, 179–82. and Fisheries 3, 242–85. Sainsbury, J.C. (1986) Commercial Fishing Methods: An Misund, O.A. (1997) Underwater acoustics in marine Introduction to Vessels and Gears, 2nd edn. Oxford: fisheries and fisheries research. Reviews in Fish Fishing News Book. Biology and Fisheries 7, 1–34. Salini, J., Brewer, D., Farmer, M. and Rawlinson, N. Moloney, C.L., Cooper, J., Ryan, P.G. and Siegfried, W.R. (2000) Assessment and benefits of damage reduction in (1994) Use of a population model to assess the impact prawns due to use of different bycatch reduction de- of longline fishing on wandering albatross Diomedea vices in the Gulf of Carpentaria. Australian Fisheries exulans populations. Biological Conservation 70 (3), Research 45, 1–8. 195–203. Samb, A. and Samb, B. (1995) Senegalese canoe fishery MRAG (1994) Floodplain Fisheries Project. Poverty, for sardinella. In: F.-X. Bard and K.A. Koranteg (eds) Equity and Sustainability in the Management of Dynamics and Use of Sardinella Resources from Inland Capture Fisheries in South and South-east Asia. Upwelling off Ghana and Ivoiry Coast. Paris: Orstom Internal Report to the Bath University Centre for Editions, pp. 362–77. Development Studies. London: Marine Resources Sech, P.A. (1980) Cataloque des engines de pêche arti- Assessment Group. sanale au Sénégal. COPACE Series 79/16. Nambiar, K.P.P. and Sudari Pawiro (eds) (1998) Tuna ’97 Smith, I.R. (1979) A research framework for traditional Bangkok: Papers of the 5th World Tuna Trade Confer- fisheries. ICLARM Studies and Reviews no. 2. Manila: ence, Bangkok, Thailand, 25–27 October 1997. Kuala ICLARM. Lumpur: INFOFISH. Stewart, P.A.M. (1987) The selectivity of slackly hung Nédélec, C. (ed.) (1975) Catalogue of Small-Scale Fishing cod gillnets constructed from three different types of Gear. Oxford: Fishing News Books. twine. Journal Conseil International Exploration Mer Nédélec, C. and Prado, P. (1990) Definition and classifica- 43, 189–93. tion of fishing gear categories. FAO Fisheries Technical Stewart, P.A.M. and Ferro, R.S.T. (1985) Measurements Paper 222. Rome: FAO. on gill nets in a flume tank. Fisheries Research 3, Panayotou, T. (1982) Management concepts for small- 29–46. scale fisheries: economic and social aspects. FAO Suuronen, P., Millar, R.B. and Järvik, A. (1991) Selectivity Fisheries Technical Paper 228. Rome: FAO. of diamond and hexagonal mesh codends in pelagic 36 Chapter 2

herring trawls: evidence of a catch size effect. Finnish M.J. Kaiser and S.J. de Groot (eds) Effects of Fishing on Fisheries Research 12, 143–56. Non-target Species and Habitats. Oxford: Blackwell Thomson, D. (1980) Conflict within the fishing industry. Science, pp. 253–68. ICLARM Newsletter (July 1980) vol. 3 (3), 3–4. Verdeaux, F. (1980) La pêche lagunaire en Côte d’Ivoire: Ticheler, H., Kolding, J. and Chanda, B. (1998) Participa- contexte sociologique et formes d’exploitation du mi- tion of local fishermen in scientific fisheries data lieu naturel. Comm. Seminaire UNESCO sur les collection, a case study from the Bangweulu Swamps, écosytèmes côtier. Dakar, June 1979. Zambia. Fisheries Management and Ecology 5, 81– Washington, P. (1973) Comparison of salmon catches in 92. mono- and multi-filament gillnets. Marine Fisheries Tregenza, N.J.C. (2000) Fishing and cetacean by-catches. Research 35, 13–17. In: M.J. Kaiser and S.J. de Groot (eds) Effects of Fishing WWF. (1989) The Footprint of Distant Water Fleets on on Non-target Species and Habitats. Oxford: Black- World Fisheries. WWF’s Endangered Species Cam- well Science, pp. 269–80. paign. Surrey, UK: WWF International. Valdemarsen, J.W. and Misund, O.A. (1995) Trawl Yamaguchi, Y. (1989a) Tuna long-line fishing II: fishing designs and techniques used by Norwegian research gear and methods. Marine Behaviour and Physiology vessels to sample fish in the pelagic zone. In: A. Hylen 15, 13–35. (ed.) Proceedings of the Sixth IMR-PINRO Sympo- Yamaguchi, Y. (1989b) Tuna long-line fishing I: historical sium, Bergen, 14–17 June 1994, pp. 135–44. aspects. Marine Behaviour and Physiology 15, 1–11. Vanden Bossche, J.P. and Bernacsek, G.M. (1990) Source Yamaguchi, Y. (1989c) Tuna long-line fishing: selection book for the inland fishery resources of Africa, vols 1 of a fishing ground. Marine Behaviour and Physiology and 2. FAO Fisheries Technical Papers 18/1 and 18/2. 15, 37–44. Rome: FAO. Ye, Y., Alsaffar, A.H. and Mohammed, H.M.A. (2000) van Marlen, B. (2000) Technical modifications to reduce Bycatch and discards of the Kuwait shrimp fishery. the by-catches and impacts of bottom-fishing gears. In: Fisheries Research 45, 9–19. 3 Marketing Fish

J.A. YOUNG AND J.F. MUIR

3.1 INTRODUCTION links and interdependencies between marketing objectives and other sectoral issues, not least being With a few exceptions, the fisheries sector has fisheries resource management and the emerging been slow to adopt the concept of marketing. His- presence of aquaculture production. torically the emphasis has been on fish stocks and production, and this broadly continues, in catch- ing what is there to be caught rather than selec- 3.2 MARKETING tively delivering what the market prefers. Even in AND MARKETS aquaculture, a primary focus has been on system capacity and production output. In most cases, dis- Though widespread in use, the term ‘marketing’ is tribution systems have grown around this supply- often misunderstood or misapplied, and it is there- driven base, with the onus on market agents to fore useful to offer working definitions. We are match consumer needs with the product available, concerned with a range of aquatic-derived food rather than to proactively determine what might products, generically described as fish or seafood. be supplied. However, with growing control and The system within which a related set of product regulation over fishing effort, allowable catches purchase decisions is made, by sellers, buyers and landings of target species, and with common and/or consumers constitutes the market. The problems of oversupply in aquaculture, it is in- marketing (or supply) chain may involve various creasingly important that producers in all fishery intermediaries in exchanges prior to consumption. sectors maximize the value of their product. Whether in developed or developing economies, Notwithstanding the deep-seated sociocultural the concept of marketing is concerned with the traditions, people go fishing not just to catch fish processes of identification, creation, communica- but also to generate income (Hannesson, Chapter tion and delivery of values. Marketing is under- 12, this volume). Neither is aquaculture just about pinned by the process of generating information the technical achievement of farming aquatic about the products and markets which are of inter- species; enterprises have to deliver benefits too. est, to enable identification and understanding of Marketing is fundamental to the realization of what consumers want and value. Having identi- such goals. This chapter explains some of the prin- fied what is valued, the next task is to create these cipal ways in which marketing can contribute to values in products to be marketed, to enable and establishing and improving the welfare gains that promote their consumption. However, in many can be generated from fishery sector production, cases, existing or potential consumers may be and considers how this might be approached in the unaware of the attributes created in products, or future. It also highlights some of the often ignored unwilling to incur time or costs in gathering 38 Chapter 3 information about these. To overcome this, and for fish are shown in Fig. 3.1, reflecting the com- achieve other promotion objectives, the penulti- plexity of the challenges involved. Within interna- mate task is to communicate with identified tional food markets capture fisheries are unique, as target markets. Having communicated with target they constitute the last major sector to rely on a buyers, the remaining task is to deliver the values hunted supply base. However, this supply increas- to the target markets, via the products created. ingly interacts with aquaculture production. The The marketing function must be ongoing and potentially destructible basis of capture stocks, cannot be effective if it is enacted only periodi- not least due to excess capital investment, and the cally, or on an ad hoc basis. It must be proactive rapid expansion of investment in aquaculture, because consumers, and the various markets that coupled with the high perishability of the product they constitute, constantly change and evolve. and the vagaries of the consumer, have combined The product attributes that individual buyers, con- to make the fish marketing environment increas- sumers and organizations value today are almost ingly dynamic and vulnerable (Fairlie et al. 1995). certain to differ from those which were valued Unlike the case for most manufactured products only a year ago and will probably be different six and services, the marketing of foods commonly months hence. The dynamics of such change are has to contend with a range of exogenous un- evident not only in the various seafood markets, certainties influencing the production process but also in the markets associated with productive (Marshall 1996). However for capture fisheries, inputs such as vessels, gear, aquaculture systems, and to a lesser extent aquaculture, marketing deci- and the raw materials used in the capture or cul- sions have to accommodate an even greater degree ture process. of external influence, arising from individual and The increasingly rapid pace of change within combined interactions of the natural and human markets, for foods or other goods and services, cre- environments. ates additional challenges for the marketing func- One instance of such interaction might be a tion. Quite apart from the time and effort required change in stock migration patterns, perhaps due to identify what markets have demanded in the to changes in seawater temperature or salinity, past, it can be much more demanding to determine which may fundamentally alter availability – and what markets currently want and to forecast their thus viability – of a product intended to satisfy future preferences. The consequent processes of various intermediaries along a supply chain. For product creation, communication and delivery example, the Western mackerel stock (Scomber also take considerable time and effort. Thus, when scombrus) in the EU Exclusive Economic Zone new products are launched they often only respond (EEZ) has varied its migratory patterns over the to yesterday’s needs rather than those of the past 20 years, swimming further offshore and moment, which will change as other firms take deeper, necessitating changes in the gear used and their own competitive actions, and as external fac- larger vessel hold capacity (Lockwood 1984, 1988; tors affect consumer behaviour. To avoid the trap Whitmarsh and Young 1985). Similarly, changes in of responding to past changes, markets must con- taste, driven perhaps by social trends, may stimu- stantly be monitored and analysed to anticipate late demand for a particular product processed emergent trends in preferences. This chapter em- from a stock hitherto underutilized, or one which phasizes the importance of the sequential nature can only now be exploited due to technical of these processes. However, before doing so, it progress in fishing gear, aquaculture techniques or is important to understand the environment in processing technology (Connell and Hardy 1982; which fish marketing decisions are made. Whittle and Wood 1992). Such events are often ir- regular, unpredictable and thus cannot be planned 3.2.1 The fish marketing environment for. As illustrated in Fig. 3.1, the importance of interdependence is also identified, whereby varia- The key elements of the marketing environment tion in one element will impact on others within Marketing Fish 39

Human Environmental Forces al Demographic Economic Tech Ethic nolog l ica ra l tu P ul o c Business Environment li io t c ic o a S ional/Quasi l l/Reg gover ona nm ti en Na ta l/ l a o n rg o a ti n a Supply Chain for Aquatic iz n a r t e Food Products i t o

n n I

Competing Producers producers

Processors Competing processors

Intermediate Competing channel intermediaries members ei c Media groups

o Markets

m Competing e retail/ r m u markets s u catering s n e i r c P a t io n l ch ia an nt O n te l c e o e l p a a and ic n Consumers: actual g o lo gr ro ap eo hic et Physical Climate change Biological M Natural Environmental Forces Fig. 3.1 The fish marketing environment. Interactions of influence and exchange: products, sevices, money and information

the chain. (For a broader discussion of such inter- tion throughout history. At various periods inter- actions of business and natural environments see national trade has both flourished and withered, McDonagh and Prothero, 1997 and Fuller 1999.) having been established on the spatial divide between points of production and consumption 3.2.2 The international markets (Coull 1972). Aquaculture technologies have also for fish created new opportunities in specific environ- ments and global locations (Paquotte 1998; FAO The global biogeography of resources has ensured 2000). Modern transportation, notably by air, has long-standing if varied patterns of fish consump- enabled a complex logistics network to be created, 40 Chapter 3 with far greater movement of product around the crustaceans and molluscs grew from 18 to 90 world. Thus for farmed Atlantic salmon (Salmo million tonnes. From a point at which only North salar), over 50% of global demand is supplied from Pacific and North Atlantic fisheries were Norway alone, much of which is marketed fresh or developed, accounting for 47% and 46% of total chilled (Liabo 1999). The resultant pattern of trade harvest respectively, areas initially noted as possi- flows is also defined by the distribution of those ble for fisheries expansion, mainly off Central willing and able to pay, and those still having fish America, Peru and Chile, in the Caribbean, off resources to trade. However, if marketing is to con- West Africa and off Australia, New Zealand, the tribute to the benefits to be attained by suppliers South Pacific Islands and the East Indies, have and consumers, it is necessary to understand the increasingly been brought into production, and main international markets and their respective into the ever-growing global trade system. sources of supply. From the major areas of traditional aquaculture An ample caveat should first be made, to under- supply in South and East Asia, an increasing line the fact that despite trends of globalization, output of high market value production has also the formation of trade blocs such as the EU Single developed. In many other regions, such as Europe, Market and other unifying forces, fish markets, as North and South America, and Australia, new for other foods, still exhibit distinctive national, industries have grown rapidly, to the extent regional and local patterns of behaviours (Gentles that major supply sources exist in almost every and Skeldon 1994; Gentles and Roddy 1999). Such continent. In 1990 aquaculture accounted for just variations may limit the applicability of wider gen- over 10% by volume of international supplies and eralizations. However, though it is impossible to some estimates have suggested that the 30% share provide here the fullest level of detail, we hope to in the early part of the new millennium may reach equip the reader with some understanding of the 45% by the year 2025 (FAO 1997; Muir and Young means by which more detailed data may be found, 1998). as is done in Section 3.3. The push towards a greater dependency on in- There are three main markets for fish: the EU, ternational trade is evident in many markets, and USA and Japan, reflecting the combination of pop- has been stimulated by overall growth in the con- ulation size, wealth and demand for diverse ranges sumption of fishery products. In 1996, 195 coun- of foods. Traditional access to supply has also been tries reported exports, and 180 recorded imports of a factor in their importance. Clearly, many other fishery products; though numbers had slightly in- significant markets are to be found, sometimes creased due to subdivision of formerly unified more disparate, and with varying and sometimes states, this confirms a substantial global inter- more specialist product demands. The former dependence. Global supply for direct human con- Soviet states, and major population centres such as sumption in 1997 was estimated at 92.5 million China and South Asia are particularly important in tonnes, some 15.9 million tonnes more than in quantity terms, and may be expected to increase 1994, exceeding the estimated rate of population their trading presence as their markets become growth, with apparent per capita availability of more open and incomes increase. Since the FAO’s food fish rising to almost 16kg. Of the 1997 pro- first estimates of world production 50 years ago, duction, some 24% – 29.5 million tonnes – were the fishery sector has developed rapidly, to the ex- used for reduction to fishmeal and oil, a ratio simi- tent that an increasing number of stocks are over- lar to that recorded in initial records. The value of exploited, few underexploited resources remain, international fish exports was estimated to have and aquaculture is contributing an increasing risen from US$17 to US$52.5 billion from 1985 to share of output and a greater share of value (FAO 1996. This was initially associated with greater 1997; Hart and Reynolds, Chapter 1, Volume 1). trade in low-value commodities such as fishmeal, During this period, the production of marine fish, reducing value per volume, but more recently has Marketing Fish 41 tended towards increased value due to higher Indeed the situation could be intractable, if only prices. Developing countries recorded a trade sur- because biological time-scales for returning stocks plus of US$16.6 billion in 1997. Fishery products to higher sustainable levels exceed the time-scales represent more than 75% of total merchandise ex- for political power. Longer-term changes in tradi- ports for Iceland, the Faeroe Islands, Greenland, tional external relationships, e.g. in ‘third-party’ Maldives and Seychelles, and between 10% and fishing agreements in Northwest and Southwest 75% of exports for a further 20 countries, including Africa, and in access to transboundary stocks such Chile, Ecuador, Kiribati, Madagascar, Mauritania, as North and West Atlantic groundfish, are also Morocco, Mozambique, Namibia, Peru and reducing EU landings. Although EU aquaculture Senegal. By value, more than half of fishery export has recorded notable gains in output, overall comes from developing countries, mainly import- supply shortfalls continue, and the EU appears ed into developed countries, and is increasingly likely to become increasingly dependent on exter- influenced by aquaculture. However, though nal sources of supplies on the international market. Thailand was the leading world exporter between The scale and composition of the growth in 1993 and 1996 (US$3.4 billion), Norway took the demand is varied, and notable differences exist lead in 1997. Farmed shrimp and salmon, respec- both between countries and within their borders. tively, have been key components. With 1997 im- For example in Germany, annual fish consump- ports valued at US$15.5 billion in 1997, Japan was tion in the northern regions is estimated at up to the leading importer, while the United States ac- 30kg per capita, compared with only 10kg in counts for ϳ10% of world fish imports. Together the interior (Young et al. 1993). In France and with the European Union (including the value of Scandinavian countries consumption averages the intra-EU trade) these three areas imported 75% around 32kgyr-1, but this is much less than, for by value of internationally traded fishery products example, the 58kgyr-1 found in Portugal (FAO (FAO 1999). 1999). While steadily rising demand can be ob- The EU represents a particularly interesting served in many economies, an exception would case of trade development given its relatively seem to be in the former Soviet states, where infra- recent creation and its diverse composition, com- structural change in supplies and relative buying bined with the supply-side factors and policy power has resulted in diminished consumption measures that have affected its source of product (Westlund 1995). (Wise 1984; Cunningham et al. 1985; Holden The overall market for aquatic food will also 1994). Traditionally, principal supply sources tend to expand through population growth (UNDP within the EU have been from capture fisheries 1997), and unless per capita income levels become in the North Sea, the North Atlantic and the severely constrained, might well widen the diver- Mediterranean, all of which have experienced sity of foods demanded. Variations in preferences declining yields over the past 20 years. In the UK, for foods can be well entrenched, even within for example, landings of cod (Gadus morhua), the individual countries and regions, and have their most popular fish species, were by the 1990s less origins in the sociocultural, psychological and than 15% of those in the early 1970s (MAFF 1999). economic context, though these in turn will Within the pelagic sector, vagaries of supply have change and will be open to influence. Just as some been no less dramatic, hindering attempts to sections of a future larger population will continue encourage consumers towards these alternatives, their current aversion to fish, others are likely to despite their acknowledged price and health maintain and expand the role of fish within their attributes. The problems of balancing sustainable diet. The dynamics of the market highlight the fishing effort with the social and economic costs need for ongoing information about the patterns of fleet reduction over a multinational shared of fish consumption and identification of what resource have defeated many policy measures. is wanted by contemporary consumers. 42 Chapter 3

tion and publication, driven by the slowest coun- 3.3 DETERMINATION try to gather and then submit returns, has rendered OF VALUES much data of value more for retrospective analysis 3.3.1 Sourcing information than for proactive marketing. However as data have become increasingly available and more up- Improvements in information technology have to-date the potential value of market databases brought about massive expansion in the ability to has increased. At the international scale a fairly identify the characteristics of different markets for sophisticated level of perspective can be gleaned fish products. This have been paralleled by the from the UN Food and Agriculture Organization need for such information to meet the demands of (FAO) (www.fao.org/fi/) and related regional competition. Databases are available with varying sites of Infofish (www.jaring.my/infofish/), Globe- degrees of access from an ever-widening range of fish (www.fao.org/fi/globe/globfish/fishery/) and providers. Notwithstanding the risks of informa- Eastfish (www.eastfish.org/). Specialist agencies tion overload and the challenges of defining and such as the Organization for Economic Co- discriminating data quality, capturing data has un- operation and Development (OECD) (www.oecd. doubtedly taken root as an essential and afford- org/agr/fish/statdata.htm) are also useful sources. able tool to assist marketing decisions (Bickerton The European Union is officially covered et al. 1998; Tapp 1998; Reedy et al. 2000). Wider through the aggregation of national statistics at discussion of the processes and techniques of Eurostat (www.europa.eu.int/comm/eurostat/) marketing research is not possible here but can Marsource (www.marsource.maris.int/) and sec- be easily accessed (e.g. Chisnall 1997; Kent 1999; toral agencies within individual countries provide West 1999; Kumar 2000; Malhotra and Birks 2000). their often more detailed perspectives at sites such Although not fisheries-specific, these texts pro- as the UK’s DEFRA (www.defra.gov.uk) or Spain’s vide a methodological template for the informa- MAPyA (www.mapya.es). A number of semi- tion sources and data from websites mentioned governmental organizations are also to be found here. To answer the characteristic questions of within most countries, providing a further layer of market research a sequence is generally followed, more detailed information. The Sea Fish Industry beginning with exploratory research and sec- Authority (SFIA, UK) (www.seafish.co.uk/), Insti- ondary data sources. Since many problems require tut français de recherche pour l’exploitation de more specific answers than are available in pub- la mer (IFREMER, France) (www.ifremer.fr/) and lished sources, a range of primary data collection Fund for the Regulation and Organization of techniques is then commonly employed. Typi- Fish and Aquaculture Products (FROM, Spain) cally these might include some combination of (www.from.mapya.es/) are typical of such organi- different types of questionnaire (structured and zations. Parallel structures are found within the unstructured), in-depth interviews, focus groups, USA; the United States Department of Agriculture observation and experimentation, depending on (USDA, www.usda.gov) and National Marine the quantitative and qualitative data required. Fisheries Service (NMFS, www.nmfs.noaa.gov) Much of this information is often already held by provide the official fisheries perspective, which is individuals through daily involvement and con- then further subdivided into regional pan-state tacts within the industry. However, to be effective, authorities. In the other major market, Japan, a systematic and ongoing approach is required to national data are available from (www.yahoo. collect and interpret the relevant data for market- co.jp/government/institutes/) and finer detail ing decision making. from (www.maff.go.jp/eindex.html) (Nakamoto At the macro scale many information sources 2000). Intergovernmental fisheries bodies such as are available, which have been improved consider- North Atlantic Fisheries Organization (NAFO, ably over recent years through their availability www.nafo.ca/), International Commission for online. Traditionally, the lag between data collec- the Conservation of Atlantic Tunas (ICCAT, Marketing Fish 43 www.iccat.es/) and International Council for the The starting point in understanding market Exploration of the Sea (ICES, www.ices.dk/) also price signals is that of first-hand sale. The initial provide valuable data. exchange in the fish marketing chain is under- Additional data are also to be found at an almost taken in a variety of ways throughout the world. At indeterminate number of trade association and one extreme, with vertical integration between private-sector sites. In some cases these may be the catching or production sector and the gathered at the aggregate international or national buyer/processor, product may simply transfer level, such as for the Federation of European Aqua- internally between corporate divisions. However culture Producers (FEAP) (http://www.feap.org/), even within a single commercial entity, distinct whilst in other cases information is provided profit centres with market-defined transfer prices to communicate a wider range of services, may be retained (Milgrom 1989). such as Howard Johnstone US (www.hmj.com/), In other instances, contracts are used to deter- Seafood Business (www.seafoodbusiness.com) mine the exchange process. These require some in (www.fishmonger.com/) and in New Wave degree of specificity about the product so that each (www.worldcatch.com). A range of more special- party is clear about what is and is not acceptable. ized consultancy services exist and in some Typically, this may suit conditions where there is cases this tailor-made research function, e.g. a greater homogeneity of supply, such as with TNS (www.tnsofres.com), or Stirling Aquaculture frozen-at-sea fish where targeted species will be (www.stir.ac.uk/stirlingaquaculture/) may prove processed onboard according to a predetermined more efficient in targeting specific needs. practice and landed to tight product specifications. Whichever site are selected numerous links are In other fisheries that are more directed and less to be found and through a process of triangula- opportunistic it may also be possible to satisfy tion the more spurious statistics can often be buyer criteria in advance. Aquaculture produc- identified or better estimates made. Little mention tion, with specific control over quality and has been made of the various hard copy and timing, may also meet such criteria. Contract more traditional information sources since sales have the advantage of reducing uncertainty these are more widely known but are also be- for both producer and buyer. Suppliers are aware coming less attractive in comparison to the data of the price they can expect for product of a online. particular specification, whilst, assuming avail- ability, buyers can be guaranteed of raw material 3.3.2 Systems of exchange to satisfy their own processing and market commitments. The need for better understanding of what is However, the vagaries of capture supply have valued by consumers within target markets is not tended to encourage the retention and refinement just confined to the post-harvest stages of the of a traditional system, which is better equipped supply chain. It is also important for producers to to deal with more heterogeneous conditions. understand what the market values and how much Auction systems of exchange attempt to equate buyers are prepared to pay, particularly as increas- market demand to the available supply of a range of ingly restrictive limits are placed on the volume different species, each of varying size and freshness and composition of what can be caught. Skippers grades, of a highly perishable product within a can use market information to make decisions short period of time. Fish auction markets repre- about when to stop fishing and land their fish, and, sent a complex exchange arena, not least because in more mobile operations, also decide where their of their dependence upon the harvesting of a fish will be landed. Likewise, for aquaculture, renewable and potentially sustainable resource, market information can feed into short-term which can also easily be destroyed. As noted production and harvesting strategies (Nagle and elsewhere, the bioeconomic characteristics of Holden 1995). fisheries have tended towards overexploitation 44 Chapter 3 and consequent wide fluctuations in supply irrespective of whether a price incremental (with (Hannesson, Chapter 12, this volume). ascending bids) or a ‘Dutch’ price decremental In different countries auctions tend to have auction system is used. their own mix of buyers, in many cases typically In either system, the intention is to identify the drawn from fish processing companies and other buyer willing to bid the most for a given lot or se- channel intermediaries, who will seek to trans- lection of fish, and a sequential selling strategy form the raw material in some way before it is sold is commonly employed to effectively skim the on. Supplies to the market may come from a market. This is achieved by selling the fish from range of sources: landings from boats based at the individual boats in near homogeneous lots. For ex- same port, foreign vessels, and additional supplies ample, the largest and freshest cod would be of- brought to larger markets by other routes. Sourc- fered for sale before moving to the next lower size ing widely helps to even out natural fluctuations in grade of cod and so on in similar fashion for each of supplies. The ownership structure of fish auction the other species. With price incremental systems, markets, like the harbours themselves, varies from the greater transparency of the bidding process, private, public, cooperative or trustee status, and whereby ascending bids are sequentially invited, each undertakes a management function reflect- should result in the eventual buyer only paying ing the interests of the parties involved. It typically one money division above that of the nearest rival involves some scale of charges being applied bidder. This is not so with a Dutch auction, be- according to the value of the fish sold, and therein cause the first to bid automatically secures the lies the incentive for auctioneers to extract the product and this may be well above the maximum highest possible price from participating buyers. of the next intending buyer. However, in both sys- Auctions are traditionally based upon an initial tems high bids are encouraged, as competitors display of the available fish so that buyers can in- have no prior knowledge of the quantity that may spect the quality and determine their procurement be taken at any successful bid level. There is al- strategy, which will also be influenced by supplies ways a risk that the highest bidder may take most likely to be available at other ports. Sales are or all of the offered supply. Buyers may also cut commonly held on a once-daily basis and are com- their margins in order to secure product and may pleted within a couple of hours, time and tempera- be willing to risk generating extra yield from a ture control being critical constraints. In effect, given lot to cover a high bid. In price incremental the supply is fixed because fishermen cannot in auctions a competitor may simply want to drive up practice return to sea and land more fish within a rival’s raw material costs and so generate a com- that same period even if favourable market condi- parative advantage. This can be achieved by partic- tions emerge, just as buyers can only deal with a ipating in the bidding process but, using market given quantity before product quality deteriorates. intelligence, intuition and a bit of luck, withdraw For these reasons, it is important that information just before the rival reaches their ceiling price. Tra- searches are undertaken thoroughly by all parties. ditionally, in display auctions, uncertainty is also If the highest bid prices are to be realized from promulgated by the secretive way in which bids are buyers it is necessary to ensure that the positive communicated to auctioneers, although with attributes of the available product are communi- automated clock systems this element has been cated to them. This market intelligence is, of course, reduced. In addition to making bid-price decisions supplemented by a network of contacts with other on the basis of information available in the market markets as the sales progress. Notwithstanding at the time, buyers must also incorporate more the available information flows, the vagaries of the opaque information about future supplies to the raw material supply chain encourage price uncer- market and anticipated demand. tainty and thus volatility in the market prices real- Inspection of fish quality is undertaken in a va- ized (Wilson 1980). Significant price movements riety of ways, ranging from the physical removal of within short time periods are commonly found, a sample of the product, such as is done with coring Marketing Fish 45 tuna in Tsukiji market in Japan, through to more smell via electronic noses exist. All have a signifi- formalized grading systems according to specified cant role to play in matching supply with demand criteria. In the EU, for example, designated species across the more accessible international market- are graded according to size and freshness criteria, place of the future. The technology already exists and boxes are labelled accordingly prior to sale. for boats to supply markets with electronic details However, this scheme owes more to underpinning of their catch as it is caught, and certainly prior to a system of producer income support, which oper- their making a decision to land. Though contract ates by establishing a floor price below which fish purchasing has been the norm in the aquaculture is withdrawn from sale, rather than providing sector, this technology also opens out possibilities product information. Traditionally, buyers have of offering product information before it is har- relied upon their own assessment of quality. vested, and opening access to a wider range of Hitherto, fish auctions have largely depended buyers. In any event, incorporating additional upon the presence of buyers in person, or proxy, at electronic product cues into the virtual market- sales, However, innovations in information tech- place will be to the potential advantage of both nology have lessened this requirement, and buyers buyer and seller since it should improve the mar- in some markets can bid remotely without ac- keting decision-making process along the supply tually seeing the product prior to purchase. Such chain. Ultimately, the ability to realize higher schemes are found onshore in Iceland, for example, prices is critically dependent upon buyers and where dispersed landings hinder the geographical consumers perceiving product to be of higher concentration of buyers. Instead they are linked value, and in turn this raises questions as to how through an online system and sales at prearranged further values might be created in the raw material times are simultaneously conducted from a of the resource base. number of locations. A rather similar concept of remote auctions is also to be found in the northeast Atlantic pelagic fisheries, where vessels will con- 3.4 CREATING VALUES duct an auction with shore plants before deciding 3.4.1 Handling and storage where to land their catch. Historically the draw- back to such remote systems of exchange has The process of creating values in the raw material been the need to know the characteristics, such as is an integral part of the marketing decision- freshness, size, texture etc. of the product being making process. This must be based on the specific sold – certainly if high prices are to be encouraged. fishery or aquaculture system, and starts with the Earlier versions of remote bidding in the 1980s method of capture or production; the selected fish- commonly suffered through the inconsistency of ing gear or the harvesting system will influence grading schemes. Product perceived to be of uncer- and/or irrevocably alter the attributes of the prod- tain description naturally tended to discourage uct landed. Gear types are discussed by Misund buyers from making high bids and so left fisher- et al. (Chapter 2, this volume), but examples may men dissatisfied with prices. Branding strategies, illustrate the complexity of the gear selection such as the ‘Silver Seal’ product of Scheveningen in decision. Many factors are at play, including: the Netherlands, which is confined to fish no more regulations, selectivity, geographical location than one day old, have been attempted, but more of the resource at different seasons, financial widespread gains are likely as technology enables viability of gear types, technical constraints and accurate product information to be sent to, and changing market preferences. verified by prospective buyers. In effect this accel- Perhaps the clearest instance of how value may erates the traditional process of establishing trust be created through the mode of capture and subse- in a particular supplier. quent handling is in cases where product is sold Digital online images are increasingly avail- alive. Such methods typically require the stock to able, and other remote sensory stimuli such as be taken with minimal stress and physical dam- 46 Chapter 3 age, and, consequently, enclosing devices such as sal fisheries, and in some aquaculture sectors, fish traps, stake nets and some forms of lines may be may be bled at slaughter to produce a whiter flesh employed (Misund et al, Chapter 2, this volume). colour, or reduce blood spots, and prolong shelf- For aquaculture, crowding and brailing devices are life. In other cases fish may be stored in different commonly employed, with specially designed sizes of containers mixed with different types of transfer pumps to holding units. Where the fish is ice, to optimize cooling conditions and hence trawled or towed for some time in a net, con- preserve shelf-life. strained too tightly, and/or subject to mechanical Whatever the origin, a key prerequisite is the shock, damage will tend to be sustained to at least exchange of a raw material that will provide a basic some proportion of the product. For pelagic fish- building block to which further values can be eries, purse seining is generally recognized to yield added. For a number of reasons, the transfer of fish superior quality product to that produced from from the boat at the point of exchange has often trawling, as the fish is enclosed rather than been identified as a weak link. Similar issues can dragged. However, trawls can be used in a much be observed in aquaculture systems. Assuming wider range of sea states and can take fish at greater fish is cooled down after capture or harvest, there depth, and may be more profitable to work, regard- can be a break in the temperature-controlled less of the potential for lower quality output. Aside environment when it is laid out for inspection by from the more obvious effects of physical damage, buyers. The significance of this break in the chilled a range of evidence suggests links between or cold chain depends on its duration, and on ambi- stress encountered during capture and subsequent ent temperatures; in some cases fish may be changes in biochemistry which may affect texture, brought from contact with ice to a market with air taste and keeping quality (Lemon and Regier 1997; temperatures of 30°C or more. Values may also be Ehira and Uchiyama 1986; Kaylor and Learson lost through additional handling of product by 1990; Perez-Villareal and Pozo, 1992). This may es- both prospective buyers and ancillary staff. This is pecially be the case in faster swimming pelagic especially so where product is stored in bulk and species, but may also apply in other fisheries, then graded according to species, size and other whose established capture methods may have quality criteria. Even where the need for such hidden the potential to improve quality. handling is reduced through boxing at sea, or on Good post-harvest or onboard handling practice harvest, physical damage often happens through is fundamental to the quality of the end product, overfilling of boxes and excessive pressure then few of whose determinants of perceived quality, if being placed upon the fish in the vessel’s hold or in any, can be restored once removed. Keeping to rec- storage areas, and possibly when stacked on the ommended practices, temperatures and time in- market. In one recent study of white fish landed at tervals between the stages of gutting, storing and the EU’s largest port, Peterhead, Scotland, it was otherwise dealing with the catch will help to found that overfilled boxes resulted in an 8% create values which can be built upon later in the weight loss and, more significantly fish from boxes supply chain. Methods may change according to not overfilled attracted prices up to 15% higher season and market demands, and at different times (Harman 2000). The use of larger containers the handling options may take on different priori- affording more protection for the fish, combined ties. For example, when abundant fish are encoun- with more accurate weighing, grading and re- tered late in a trip the skipper might decide to cording of where and when particular lots were concentrate on catching more fish and landing captured or harvested, is likely to become more them round rather than spending time gutting. prevalent and should help to reduce the loss of Innovations in handling the large volumes of fish value at these stages. Such measures of product found in pelagic fisheries have resulted in fish traceability are central to the concept of Hazard pumps being adopted to transfer fish from the net Analysis Critical Control Points (HACCP) and into the hold and to land onshore. In other demer- are likely to become a prerequisite of more and Marketing Fish 47 more marketing channels, including downstream the retail food market where products have to processing. compete with an ever-widening array of fish and non-fish substitute products. 3.4.2 Processing methods and Critical to the success of the contemporary market influences processed fish product has been the incorporation of attributes driven by consumer wants within the Fish processing is essentially concerned with the wider market for foods. Creating products with transformation of a raw material into an almost values which are perceived to be important by the limitless array of products offering a range of differ- consumer, rather than simply because processors ent values. Historically, processing was under- have the technical ability to include them, is a sig- taken to preserve fish to enable its storage and later nificant determinant of product success. Effective consumption when alternative protein sources marketing necessitates understanding the posi- were not available. The evolution of different pro- tion a value occupies within the mind of the cessing techniques, notably salting, drying, smok- consumer, and why it is located there. Having ing, freezing, canning, chilling and other methods gathered the information about the competitive is well documented (Cutting 1955; Burgess et al. position of the product attributes within the 1965; Aitken et al. 1982). More recently, fish- market, and a clear understanding of what con- processing technology has been to the fore in the sumers perceive to be important, decisions can wider arena of food science and technology, not then be made as to how these attributes might be least because fish is one of the most perishable created within the product range. foods and has therefore generated very demanding Consumer perceptions of fish, with some ex- technical requirements. Fish has been used suc- ceptions, have generally become more favourable cessfully in a variety of technical innovations in over recent times. In many countries, a notable processing and so appears in an extensive range of trend has been the increased concern with healthy products (Connell and Hardy 1982; Young et al. eating, in which fish is perceived very favourably. 1992). Product diversification is manifest in the For many white fish species, a frequently cited increasing number of new products, some using point is the low level of fat, while for the typically different species, continually entering the market. more oily pelagic species, high levels of omega-3 New product development has evolved from polyunsaturates may confer important benefits the simplest levels of whole fresh fish, through against cardiovascular diseases (Marshall and progressive increments of gutted, filleted, deboned, Currall 1992). Health concerns have been further steaked, and otherwise modified raw material, and stimulated through food scares, especially with through processes such as smoking, marinating, BSE (bovine spongiform encephalopathy) in the adding sauces and the inclusion of other non-fish UK, where high-profile media attention has en- foodstuffs to add value. The differentiated bundles sured more widespread coverage and greater suspi- of attributes thereby built into the products may cion of traditional foods (Smith et al. 1999), leading incorporate additional intangibles such as conve- to increased interest in the potential safety and nience in storage, in preparation, and in consump- nutritional benefit of foods such as fish. However, tion features. Packaging in a widening array of despite some gains, it must be noted that this more options including chilled, vacuum skin, and modi- favourable disposition has not eliminated many of fied atmosphere packs, frozen packs, cans and the traditionally negative perceptions harboured pouches also provide the opportunity to further by potential or actual fish consumers. Although differentiate the end product. The pack format, negative perceptions vary internationally, reflect- shape and presentation, and in particular, the op- ing the diversity of social and cultural influences portunity for labelling and branding, all provide an which have shaped broader patterns of food con- important channel to communicate further with sumption, their continuance remains a significant the consumer. This is especially significant within barrier to the wider acceptance of fish, generically 48 Chapter 3 or as specific products, amongst different user social or economic importance of the industry to a groups. Consumers’ perceptions are also prone to particular locality. More commonly, support may negative influence, and, in the UK, publicity sur- be derived from some form of taxation or levy paid rounding a single television programme about the on the product purchased or sold, and funds raised alleged toxicity of farmed salmon and its environ- may contribute to the identified promotional ob- mental impacts fuelled a widespread media cover- jectives. Many such organizations also serve as the age berating the industry (see, for example, Millar principal centralized source of market informa- 2001; Alba and Dennis 2001). Differences in the tion and may be accessed by other firms operating values placed on fishery products by different within those markets, as discussed in Section 3.2. groups, and their susceptibility to influence, has On the international scale, FAO Globefish and the in turn stimulated a range of organizations to at- various regional divisions of Infofish, Eastfish, In- tempt to communicate the values of fish more fosamak, Infopesca and Infopeche cover the widest effectively to consumers. remit. Elsewhere, various national and regional or- ganizations also operate, e.g. Fonds d’intervention et d’Organisation des Marches des Produits de la 3.5 COMMUNICATING Pêche Maritime (FIOM) in France, SFIA in the UK VALUES and Alaska Seafood Marketing Institute (ASMI) in Alaska, USA, each with varying levels of coopera- The product and all the attributes created within tion and sharing of resources. it present a potentially powerful mechanism to Communication of product values occurs in a communicate signals to buyers and consumers. multitude of ways and is by no means confined to However, sellers cannot afford to rely upon advertisements; in many cases potential buyers spontaneous purchase or consumption of prod- are not consciously aware of their active reception ucts, because these acts themselves first require of signals sent. Each day individuals may be sub- individuals to be aware, and perceive the value, of ject to hundreds of signals, many of which will the products, often against competing messages simply pass with little, if any, recognition. Else- from alternative products. Informing individuals where, for example in a supermarket, the shopper about products involves a range of promotional will encounter brands and other point-of-sale sup- tools and a variety of media, which are discussed port which will communicate something of the more generally elsewhere (Cooper 1997; Shimp attributes of the product, and may encourage (or 2000; Schultz and Kitchen 2000). Advertising, discourage) a subsequent decision to try the prod- publicity, sales promotion, personal selling and uct. Branding is a pervasive tool, employed to direct marketing, increasingly via e-commerce, all achieve a number of objectives, including product have specific advantages and disadvantages in differentiation, price differentiation, increased different situations, and a combination of these brand (product) awareness, customer loyalty and will commonly be employed to achieve commu- repeat purchasing. It can be done at a variety of nication objectives. levels ranging from the generic, covering one or a The promotion of fish is undertaken by a range range of species, possibly from a particular geo- of different organizations around the world, each graphical area, or through the product range of a with more specific remits and organizational particular retailer or caterer. It may also be applied structures. For some organizations, promotion, to just some part of the product range of a fish often primarily advertising, is the sole activity, processor, possibly a premium-quality product, whilst others have a more eclectic industry focus enhancing price and image, or at the other end of which may, for example, include post-harvest the spectrum, a low-price product, encouraging work, gear technology and so on. Organizations are volume sales, and allowing margins to be retained also funded in a variety of ways: some receive in price-competitive conditions. public-sector grant support on the basis of the Over recent years, and especially within aqua- Marketing Fish 49 culture, more use has been made of branding to Promotion strategies are not necessarily con- indicate attainment of a given quality standard fined to increasing trade; attempts have also been defined by meeting specified criteria. Examples in made to influence the type of fish eaten by target the farmed salmon market are numerous, such as population groups. For example, in the 1990s a the Tartan Quality Mark for Scottish salmon, number of organizations responsible for market- some of which also enjoys the distinctive award of ing fish in various EU member countries cooper- the French Label Rouge mark, a standard synony- ated to communicate the benefits of increasing mous with high quality in a range of foods (Lloyd consumption of small pelagic species. In the 1997 1999). Such brands are important because they im- transnational ‘European Fish Campaign: Health mediately serve as a communication tool to the and Wealth from the Sea’ younger consumers were buyer, and potentially the consumer, that the prod- targeted with health messages (high omega-3 uct has attained a certain guaranteed minimum polyunsaturates) to promote the consumption standard. This standard may of course be surpassed of sardines (Sardina spp.), herring (Clupea spp.), by yet higher quality specifications being applied, mackerel (Scomber spp.) and sprats (Sprattus spp.) at which point the brand would then simply seek (Seafood International 1997). In addition, stimula- to reinforce superior status upon the base creden- tion of the domestic EU market would reduce de- tials of the generic mark. Such generic standards pendence upon export markets and, in comparison may have the problem of being vulnerable to the to white fish, the pelagics are cheap. The decision adequacy of the monitoring and enforcing system to focus upon a particular message is again de- or organization – a task not so easily accomplished pendent upon the need to identify and understand uniformly throughout the myriad of supply what existing perceptions of the product are held, chains. In the market, products are liable to be and by which segments of the market, using a judged by buyers according to lowest guardian of range of demographic criteria, so that the commu- the standard, and there is thus the potential for one nications campaign can be designed accordingly. member of the brand cartel to damage the reputa- Elsewhere, communications campaigns have tion of other producers. sought to limit consumption of certain fish prod- Campaigns specific to species and/or country or ucts. The notable cases of dolphin-friendly canned location of origin have been the focus of increased tuna and the 1990s campaign by some US chefs not communications effort over the past years, and es- to serve swordfish on their menus, are discussed in pecially as international trade in fish has grown. In Section 3.7.2 and highlight the often ignored role Norway, the fifth largest producer of fish in the of communication in effecting resource manage- world, but with a population of under 5 million, a ment measures. Another instance is the Interna- considerable effort is placed upon promoting the tional Whaling Commission (IWC) ban on trading positive attributes of Norwegian aquatic food whale products – yet within Japan and Norway products. Estimations of the effects of promotional these products are still promoted to the consumer spend on different fish market sectors have been in much the same way as other competing foods. undertaken by a number of academics, notably Such communications, for and against, different within econometrics (Bjorndal et al. 1992; Asche products and capture methods may be seen as and Sebulonsen 1998; Kinnucan and Myrland an attempt to manage fisheries through demand 1998, 2000). However, in the real-world marketing rather than the traditional supply-side measures. environment wherein these campaigns run, com- This interaction between marketing and fisheries munications are subject to a multitude of con- resource management is increasingly important temporaneous influences which range from and is returned to later. fluctuating and competing prices, to the weather, Individual manufacturer brands of fish prod- and make this a complex and difficult task. Most ucts should also be recognized as an important part certainly, it is not a scenario where a given spend of the communications process, whereby the same can guarantee some certain return in the market. fundamental task of getting some particular mes- 50 Chapter 3 sage across to the target market is attempted. In Export Council in promoting the utilization of some instances, these have been undertaken on an salmon by chefs in China; a species that hitherto international basis, such as Bumble-bee tuna and had remained little used in this important ethnic the infamous Captain Birds Eye, who has the dis- cuisine (Chen 1999). Such routes are potentially tinction of rebirth, having once been drowned by invaluable in the case of new species or products, his creator Unilever. Otherwise, manufacturers’ especially where they are unfamiliar, simply be- brands are more localized. Within the retail sector cause in the catering situation, the consumer is pre- an important trend has been the incorporation of sented with a product solution rather than a new retailers’ own brands into the product range, a phe- task to be overcome (Maddock and Young 1995). nomenon that owes much to the growth and domi- Of course, in many markets for fish the product nance of supermarket chains in many countries, is still treated as an undifferentiated commodity, as discussed in the next section. Supermarkets and each transaction is simply based upon the have moved from simply being stockists of a product cues perceived from the product inspec- manufacturer’s brand, to a position whereby the tion at that time, rather than any others emanating supermarket’s name becomes the brand, and the from supporting experience or explicit promo- statement about the products to be found therein tional activity. Communications may confer (McGoldrick 1990; DeChernatony and McDonald added values but they can also represent a cost that 1992; Burt 2000; Walters and Hanrahan, 2000). in some situations is less affordable and not justifi- This change has led supermarkets to become more able. Whilst even in the poorest countries, brands, closely involved in the whole process of develop- advertising and related promotions activity may ing products, often in close conjunction with be present, communications do not necessarily manufacturers. Such codeveloped brands are often have to involve large promotional budgets. Al- able to command a premium position within the though they often do, and especially where wider- market and attract a commensurate price and reach advertising is employed, increasingly more profit level. direct forms of communication are both possible Communications about fish within the cater- and effective. This has been encouraged through ing sector are often more concerned with reaching the development of the Internet and the more gen- the buyer or user rather than the consumer. eral trend to e-commerce. Individual manufactur- Although there are some exceptions where a strong ers and other industry organizations now have fish brand is present in the away-from-home access to a rapidly expanding audience with whom market, such as Red Lobster restaurants in North they can communicate in detail and individually; America, or Harry Ramsdens fish and chip restau- this was simply not possible even just five years rants in the UK and elsewhere, fish has not ago. Initially it has been the higher unit value, approached the levels of awareness encountered speciality-orientated products that have tended to in other fast-food and catering sectors such as be promoted on the net. Such products commonly McDonalds (Ritzer 1993). Whilst end-consumers encounter barriers to market entry because of their may not necessarily be the initial target of commu- more specialized market volumes, higher prices nications strategies within the catering sector, and hence lesser ability to meet the more standard other channel intermediaries are. Buyers, and users requirements of retail chains or many catering out- such as chefs, represent a potentially valuable con- lets. Notwithstanding the current reappraisals of duit through which positive messages may be sent web-based marketing and sales businesses, the to diners, and all the more powerfully since com- promotion opportunities afforded by the Internet munications will be delivered to their stomachs. and the evidence of many individual websites op- Chefs and other channel intermediaries may pro- erated by some of the very smallest of operations vide the opportunity to encourage adoption of new suggests that this channel of communication will products and different styles of preparation. This also increasingly alter the conventional concept of was the focus of an approach by the Norwegian delivering the product. Marketing Fish 51

needs of the more powerful buyers. This is espe- 3.6 DELIVERING VALUES cially evident in the case of supermarket chains, 3.6.1 Cold and chill chains where their individual group-buying capacity is such that they purchase direct from the fish The delivery of values within the fish marketing processor or importers for much of their product chain is arguably more complex than in any other range rather than through intermediaries. How- food product sector. Logistics have to cope with an ever, a number of such large organizations still uncertain supply, which often cannot be con- obtain certain more specialist parts of their firmed until within a few hours of transportation, product range from wholesale markets. to variable locations around the globe in a dedi- Inland wholesale fish markets also remain im- cated temperature-controlled transport environ- portant for smaller and often specialist fish retail ment within the minimum journey time. The outlets which need comparatively small volumes, comparatively rapid rates at which fish products and wish to secure a selection from a number of dif- spoil places a premium upon more rapid transport ferent suppliers. Many parts of the catering sector systems, which cannot always be justified by the may also rely on independent outlets within the unit value of some product categories. This is per- wholesale market. As might be appreciated, highly haps taken to extremes in the case of small pelagic complex supply chains soon evolve within such products, which combine high endemic rates of markets where there are many independent buyers perishability with unpredictable loads that can from a number of specialist traders. Complexity is quickly amass very significant volumes and yet heightened by the typically restricted hours of still command only limited prices on the market. trading, which result from a perishable product For example in 2000 the ex-vessel price for herring that needs to reach the next stage in the chain with in the UK was under 10% that of cod, yet both the minimum of delay, often reaching the con- products would incur similar unit transport costs sumer within the normal working hours of that (SFIA 2000). Because of these fundamental charac- day. Emergent demands for Hazard Analysis and teristics the transportation of fish, especially fresh Critical Control Point (HACCP) systems, and the and chilled product, has increasingly tended to need for compliance with increasingly stringent favour road and air distribution rather than rail. health and hygiene regulations have contributed However, on intercontinental routes frozen, to quite radical revisions of many market struc- canned and dried products are also carried by sea. tures. However, in a number of instances, tradi- Marketing channels within the fisheries sector tional supply chains have lagged behind the have evolved from the traditional network of port standards demanded by buyers, resulting in their wholesale markets and inland wholesale markets exclusion from emerging market developments. feeding a widening succession of smaller inter- This has been especially so with supermarkets mediaries and buyers. Like port markets, inland and so has accelerated more efficient logistics, wholesale markets arose because of their ability whether these are managed over a network of to provide buyers with a wider and more con- regional warehouses or through integrated centrated range of product from a number of temperature-controlled infrastructures at the port different supply points, so lessening the impact of market. supply fluctuations. Whilst a number of wholesale markets remain important, e.g. Rungis in Paris, 3.6.2 Fish retailing Merca in Madrid, Tsukiji in Tokyo, they typically now handle a lower volume of product, often from Food markets have become far more internation- a more diverse range of supply sources, and cover a alized over the relatively recent past, and this has wider range of species. As the market structure of resulted in brands being available in many differ- the food industry has become more concentrated, ent countries, and a much wider range of compet- so too has the supply chain changed to meet the ing products being available within most markets. 52 Chapter 3

Changes in individual patterns of consumption, ration of traditional specialist outlets within their both at home and outside, have fuelled develop- layout. Especially within some of their more re- ments within the food retailing and catering cent smaller-scale operations, supermarkets have sectors. In developed countries especially, super- also been able to benefit from top-up shopping markets have gained dominance in the packaged trips, often made by consumers willing to pay a grocery market within a relatively short space of premium for quality convenience products. In time. For example, in 1973 only half of UK pack- Germany, Ireland, the UK and France the super- aged grocery sales was through supermarkets, market sector now accounts for between 30% and whereas 20 years later this was typically over 80% 70% of the fresh fish market (Gentles and Roddy (Burt 2000). In the late 1990s the three leading multi- 1999; SFIA 2000). ple chains in the UK, Germany, the Netherlands and Belgium all held packaged grocery market 3.6.3 Fish catering shares of over 40% (Gentles and Roddy 1999). Fish has followed this trend more slowly, being one of In broad terms the diverse catering market can be the last fresh food categories to be dominated by split into two categories: the profit and cost sec- supermarkets. Evidence indicates that this is be- tors, the former representing hotel, restaurant, cause of lower comparative profitability in this café and take-way outlets, and the latter larger- product line, and the product-specific retail skills scale typically institutional units such as schools, required (Young et al. 1993). However, the incorpo- hospitals and canteens. Consumption of fish with- ration of fish products, and especially fresh fish in the catering sector is influenced by more general displays, may contribute positively to the store’s trends in food consumption away from home. ambiance as a supplier of fresh produce, and super- Consumer uncertainties about product storage, markets are increasingly extending their domi- preparation and odours associated with cooking nance in these sectors. Although there has been a and/or consumption can be reduced or removed, dramatic reduction in the numbers of specialist and consumers can be expected to become more fish retailers, a significant residual core remains in adventurous. The importance of food consump- most markets. Typically, these have been able to tion away from home varies significantly in terms exploit niche markets, often based on a competi- of expenditure and in its role within society and tive advantage of service, product knowledge, culture (Warde and Martens 2000). Food consump- product range and quality. tion away from the home accounts for 30% of More positively, the presence of fish within the household food expenditure in France, Germany supermarket has increased access to potential con- and the Netherlands, whereas in the UK this sumers, many of whom would not deviate from amounts to only 20% and this level is only half of their shopping pattern specifically to purchase that in the USA (Euromonitor 1995; Key Note fish. Supermarkets have helped reintegrate fish 1994, 2000). However, within many market sec- product options within the less frequent shopping tors a far higher proportion of expenditure will be pattern of the food consumer. They have long been devoted to eating out, especially at more extreme leaders in frozen and canned fish sectors, common- ends of income brackets where individuals range ly holding 60–80% of the market, because the from being wealthy but time-poor to those with- longer shelf life of these product forms is more out home or cooking provision. Within both the suited to less frequent shopping. Before the wider profit and cost sectors there are many individual adoption of packaging innovations such as modi- players, some with a number of outlets. This rather fied atmosphere packs (MAP) and vacuum skin disparate and disaggregated industrial structure packs (VSP), the shelf life of fresh products was is significant, since smaller outlets present lower often too short for the shopper’s buying cycle. entry barriers to suppliers and thus provide More recently, the involvement of supermarkets opportunities to access fish consumers which with fresh product has increased through incorpo- may otherwise be denied through supermarkets. Marketing Fish 53

Nationally, the relative importance of the catering facilitate access to marketing information market for fish varies: typically amounting to just systems and now offer far greater scope for more under half the market in the UK but in France, Ger- market-based fish-capture decisions. Individual many and the Netherlands remaining at around skippers can access market information and use 30% (Key Note 2000; SFIA 2000). this proactively to target their quarry, in terms of species, size and subsequent onboard handling decisions, with greater certainty than hitherto. 3.7 FUTURE VALUES Improvements in information systems may also facilitate the application of resource manage- Having described the ubiquity of marketing phe- ment measures. Remote detection of the volume nomena in food consumption, and considered and composition of what has been caught is theo- some of the more established ways in which mar- retically possible with existing technology (e.g. keting issues are manifest for fish products, it is Strachan et al. 1990) and it is not a great extension useful to outline the ways in which marketing of current trends to foresee a regulatory system might contribute to the fisheries sector in the fu- whereby remote monitoring of catches will be- ture. The tendency to overcapitalization in fish- come routine. In aquaculture, increasing concern eries and the consequences of excess fishing effort for production and for environmental manage- have been discussed elsewhere (Hannesson, Chap- ment may give rise to similar applications in mon- ter 12, this volume). The aquaculture sector has itoring and verifying processes, and in attesting also grown significantly, and contributes a grow- that specific outputs have arisen from approved ing proportion of high-value products, particularly systems and management operations. Whilst in the fresh and high-quality value-added sectors. there may be reservations concerning the imposi- It has also contributed increasing inputs to devel- tion of various monitoring processes to remotely oping country economies and their poorer rural identify species, production or capture methods, and urban populations, and in some cases has and to identify and quantify the output, the oppor- delivered important export earnings. In doing so, tunities to facilitate enforcement, hamper false it uses a range of resources and may have less declarations and support wider consumer confi- desirable impacts (Naylor et al. 2000). Overall, dence have definite advantages. however, the role of marketing in resource man- The effectiveness of resource management agement and technical development has tended measures may be undermined by the understand- to be overlooked, both in fundamental issues, and able attempts of fishermen to extract the maxi- in the various policy instruments devised to mum value of their quota entitlement. In mixed address these. fisheries it is often impossible to target single species without capturing others, and bycatches 3.7.1 Marketing and can become a significant problem (Kaiser and Jen- resource management nings, Chapter 16, this volume). The discarding of species, sometimes perfectly edible but simply The role, and the potential, of market information less valuable in that market, introduces a loss and systems in influencing fishermen’s behaviour was represents an additional cost to both the fishermen noted earlier in Section 3.2. Because fishermen go concerned, through time and effort in clearing to sea to earn a living, decisions about target their gear, and to society at large who might other- species, where to land, and price expectation, are wise have used the ditched resource. Ecosystem inextricably linked to marketing. The tighter regu- impacts, though difficult to specify, may also be an latory environment in which producers fish re- issue. Even where catches are well targeted, quota quires them to extract the greatest value possible restrictions will tend to encourage the selection of from their allocations. Improvements in ICT the highest grade sizes (MacSween 1984). High (information and communications technology) grading can result in more juvenile stock being 54 Chapter 3 returned to the sea, often dead or with very little 3.7.2 Green values chance of survival; again representing a cost. Regu- lations which preclude discarding or high grading In addition to the marketing implications in suffer from the classic problem of enforcement managing fisheries resources, a related issue is the cost. Whilst certain types of equipment may be ‘greening’ of consumers. The emergence and im- banned, for example onboard grading machines in pact of green values amongst consumers varies in pelagic fisheries, or various gear adaptations regu- different markets. However, even where consider- lated in size and/or configuration, in many fish- able disparities exist between nearby countries, as eries practical enforcement is at best non-uniform. is the case in some EU member states, the tenden- Interactive onboard monitoring may provide some cy is for heightened environmental awareness, scope for improvement but areas of interpretation manifest in various ways (Wessells 1998; Aarset et may still cause difficulties. For example, the point al. 2000). Concerns with more sustainable con- at which a skipper is deemed to discard a catch may sumption are evident from markets for consumer range from the point at which it enters the net, to durables such as washing machines and fridges its arrival aboard. The impact of stress induced by through to those for fast-moving consumer goods interaction with the fishing gear may alone be suf- including food and fish products. Whilst consumer ficient to cause mortalities. It is arguable whether concerns for responsible consumption of wild improved market information systems will reduce species have been apparent so far in only a compar- such problems – for example by improving the re- atively limited number of cases, once the implica- turns available on more costly but better targeted tions of consumption are established, awareness fishing – or exacerbate them – by increasing the will also become heightened in respect of other temptation to high grade for spot opportunities. species and/or stocks. Similar trends may also be Remote electronic monitoring and recording observed in aquaculture (Muir et al. 1999). The pre- systems should, however, be able to provide some cision of the issues involved may be questionable solution to the practice of misrepresenting fish at times, and may represent a mix of concerns for landings. In the latter part of the 20th century the conservation, sustainable management, impacts false declaration of fish landings emerged as a sig- on bycatch, use of critical resources, social and nificant problem in a number of fisheries as quotas environmental impacts, contamination risks and became tighter (Ahoven 1998; Johnston 2001). food safety (Wessells and Anderson 1995; Wandel Supplying fish through illegal channels, com- and Bugge 1997). A classic example of concerns monly referred to as black fish or grey markets, with conservation, and latterly sustainability, re- simply undermined the clarity of market signals lated to dolphin mortalities associated with tuna elsewhere and represented another instance of the fisheries. In response, and broadly in line with need to incorporate marketing actions within more formal management regulation, processors the processes of resource management decision have nurtured and widened the market for making. Remote recording of catches as they are dolphin-friendly canned tuna through monitor- taken aboard would provide a more accurate mech- ing fishing methods and communicating this anism to monitor the uptake of quota from partic- attribute on the product label. In another initia- ular stocks, a factor that may be of value in tive demonstrating concerns with sustainability fisheries subject to intensive effort and where the of fish populations, some chefs in the US removed biomass is critically balanced. Logging catches can overexploited species, including swordfish, from obviously be fed into market information systems their menus in the late 1990s in response to declin- to improve forecasting of supplies as they become ing stocks (Seafood Business 1998). More recent available, and, as earlier noted, such information heightening of awareness may be found in the has the capacity for remote auction systems or deepwater fisheries to the west of Scotland where other transaction mechanisms to be conducted the withdrawal of the UK’s claim to territorial whilst the boat is still at sea. waters around Rockall in 1998 encouraged inter- Marketing Fish 55 national fishing effort. The species targeted are of aquaculture has the potential to provide an alter- extremely slow growing and highly vulnerable native to many existing overexploited stocks, to overexploitation (Gordon and Hunter 1994), there is a need for a fuller appreciation of the intri- raising some concerns amongst consumers about cacies of interdependencies between trophic the ethics of buying such products. A number of levels. The dependence of many high unit value examples may be found in the aquaculture sector, (HUV) carnivorous species upon feeds produced such as the concerns about social and environ- from other, sometimes overfished, low unit value mental impacts of farmed tropical shrimp produc- (LUV) stocks will not sit readily within some green tion (Bundel and Maybin 1996; Gujja and Finger- consumers’ ethical checklist (Tacon 1994; Naylor Stich 1996), and the environmental impacts and et al. 2000; see also Pauly and Christensen, Chap- welfare issues involved in the intensive culture of ter 10, this volume). However, determining that salmon and other fish species (Mackay 1999; Young the source feed materials for aquaculture product et al. 2000; Alba and Dennis 2001; Millar 2001). come in turn from sustainable fisheries and/or Greater awareness of the importance of sustain- agricultural production arguably calls for a level of able fish populations and related environmental awareness and commitment that is beyond many concerns might well become more pervasive once current consumers. Nonetheless, consumers in- the actions of the Marine Stewardship Council creasingly expect that such issues are considered (MSC) begin to send regular signals to the markets. by supermarkets’ buyers, and that they can trust This joint initiative started in the mid-1990s, join- such outlets to honour their concerns. Super- ing the international operators Unilever and the markets, in turn, are increasingly prepared to seek environmental charity World Wide Fund for Na- competitive advantages in being able to offer such ture (WWF) in common action. A central concern reassurance. is to promote only products harvested from stocks As in aquaculture, the concern of the contem- deemed to be managed in a sustainable way. By porary consumer for animal welfare has been fairly 2000 only three fisheries – Thames herring, muted in respect of captured fisheries products, Australian rock lobster (Jasus novaehollandiae) and certainly when compared to agricultural sec- and Alaskan salmon (Oncorhynchus spp.) – had tors such as poultry and veal. Thus far the very fact been approved, although it is intended that more that fish tends to remain out of sight under water will be added to as other fisheries prove their eligi- until presented, already slaughtered for consump- bility (Peacey 2000). Explicit communications tion, may have contributed to its lower profile in with consumers through media and other direct the green consumer’s sights. However experience channels, such as packaging and ecolabelling of from a number of other ‘ecowars’, e.g. offshore oil products, are liable to heighten awareness of un- exploration, demonstrates the implausibility of derlying issues including the overexploited status any attempt to cover up. Conduct such as the of many of the stocks currently prosecuted for fish selection of only the most valued parts of cap- supplies (OECD 1991; Norberg 1996; Donath et al. tured animals can expect to attract increasingly 2000; Jaffry et al. 2000). Combined with other adverse publicity. Thus the removal of crab claws activities, such as the FAO Code for Responsible and the discard of the live animal, or the finning of Fisheries (FAO 1995), it seems probable that fish live sharks at sea with the body dumped to drown, consumers will not be able to avoid being more have the capacity to engender widespread retalia- aware of the impact of their consumption decisions. tion against fisheries generally. Yet shark finning The depth of consumer understanding of the is permitted in all but four countries of the world impact of consumption remains variable, and even (Guardian 2000). Like all other food production amongst the more committed groups some systems, the fisheries sector will need to pay research suggests that quite considerable gaps in increasing attention to the importance of sus- knowledge exist (Aarset et al. 1999). Whilst super- tainable and publicly acceptable management ficial consideration may suggest that development practice. 56 Chapter 3

Aquaculture has already experienced close con- locations, in developed and developing countries, sumer scrutiny on a number of different fronts. marketing has emerged as a value-driven activity The simple proximity of aquaculture develop- increasingly focused upon raising the unit value of ments in the coastal zone and their inevitable an increasingly scarce global resource, or of an conflicts with other users alone have provoked aquaculture output reliant in turn on scarce and attention. Growing concerns with the environ- increasingly expensive resources. In conjunction mental impact of aquaculture sites, especially with apposite resource management measures and where these are shared with other indigenous implementation, this emphasis upon deriving groups such as fishermen, coastal farmers, and more value from the finite resource base could gen- with divers, tourist and recreational users, will erate positive social and economic gains for those demand more demonstrable controls over the who are involved in the fish industry supply chain, sector’s activities (Clark 1992). Similarly, con- and the adjacent often highly dependent regional ventional genetic manipulation of stock may economies. However, the possession of the skills, be associated in the minds of consumers with resources and technologies available to develop transgenic modification of foods elsewhere, and in value from the fishery and aquaculture resources Europe at least there is widespread antipathy to may become increasingly concentrated within buying such products. Concerns pertaining to the larger food industry players. If the interests of the use of chemical treatments and pharmaceuticals many small-scale operators in the current supply on the fish stock have also drawn adverse compari- system are to be maintained, their access to mar- son with activities in modern intensive agricul- keting capability and their successful application ture and will not enhance the product position in of relevant approaches may be an important deve- the consumer’s mind (Aarset et al. 2000). The re- lopment issue. cent interest in using organic principles for aqua- Constructing values within contemporary fish culture (Young et al. 2000), the technical issues products has been shown to involve a range of involved, and the communication of associated activities that are certainly not just confined to qualities to consumers is an interesting, though post-harvest processing and selling. Successful complex, market response to these and related marketing of fish products is increasingly driven concerns. This in turn has raised concerns for how by an ongoing, proactive and interdependent set of captured fish product might also acquire more posi- activities that constantly seek to deliver in re- tive green attributes, and promote these to better sponse to market signals. Many of the values incor- effect (Belknap 2000). Notwithstanding such de- porated in the product sold are intangible, such as bates, however, food products based upon aquatic the psychic income derived from consuming a resources, cultured or captured, have the potential product that has been captured or produced in an to send more favourable messages along the food environmentally positive way. In other instances marketing chain than many terrestrial food pro- values may be much more similar to those offered ducts. However communicated, these intangible with other contemporary foods, such as conve- aspects of fish products are likely to figure more nience of storage or ease of preparation. The ability prominently in the future. of the marketing function to incorporate and re- spond to such signals has increased significantly through ICT, and with developing food technolo- 3.8 CONCLUSIONS gy. The capacity of aquaculture to deliver more regular supplies of uniform product, adapted to The role of marketing within the fisheries sector meet downstream process specification, also of- has evolved quite dramatically over the relatively fers significant opportunity. Even with existing recent past. In some parts of the world today the levels of technology, let alone those on the verge of remnants of a production-based, supply-driven launch, considerable potential exists for further orientation are still to be found. But in many more market development and value creation. Whilst Marketing Fish 57 this prognosis may seem optimistic, competing Bickerton, P., Bickerton, M. and Simpson-Holly, D. food production sectors have access to exactly the (1998) Cyberstrategy – Business Strategy for same opportunities. The challenge remains for Extranets, Intranets and the Internet. Oxford: Butterworth Heinemann. aquatic food producers to take advantage of change Bjorndal, T., Salvanes, K.G. and Andreassen, J.H. (1992) to promote and develop the many positive features The demand for salmon in France: the effects of mar- of their products, and to use the benefits derived to keting and structural change. Applied Economics 24, support the sustainability of their sector. 1027–34. Bundel, K. and Maybin, E. (1996) After the Prawn Rush: The Human and Environmental Costs of Commercial Prawn Farming. London: Christian Aid Report. REFERENCES Burgess, G.H.O., Cutting, C.L., Lovern, J.A. and Waterman, J.J. (1965) Fish Handling and Processing. Aarset, B., Beckmann, S., Bigne, J., Beveridge, M., Edinburgh: HMSO. Bjorndal, T., Bunting, M.J., McDonagh, P., Mariojouls, Burt, S. (2000) The strategic role of retail brands in British C., Muir, J.F., Prothero, A., Reisch, L.A., Smith, A.P., grocery retailing. European Journal of Marketing 34, Tveteras, R. and Young, J.A. (1999) The EU consumer 875–90. and the polysemy of ‘organic food’: consumer views on Chen, A. (1999) China as market for salmon. Paper pre- regulation and definition. In: A.N. McAuley and L. sented to Salmon Summit ’99. Copenhagen, Denmark, Sparks (eds) Proceedings of the Academy of Marketing September. Eastfish. Annual Conference: New Marketing, New Relevance: Chisnall, P. (1997) Marketing Research, 5th edn. London: Evolution and Innovation for the next Generation McGraw-Hill. Stirling, pp. 287–308 (unpublished report). Clark, J.R. (1992) Integrated management of coastal Aarset, B., Beckmann, S., Bigne, J., Beveridge, M., Bjorn- zones. FAO Fisheries Technical Paper, No. 327. Rome: dal, T., Bunting, M.J., McDonagh, P., Mariojouls, C., FAO. Muir, J.F., Prothero, A., Reisch, L.A., Smith, A.P., Tvet- Connell, J.J. and Hardy, R. (1982) Trends in Fish Utilisa- eras, R. and Young, J.A. (2000) Technical deliverable tion. Farnham: Fishing News Books. RTD 5: assessment and prospects of the production Cooper, A. (ed) (1997) How to Plan Advertising. London: and consumption of organic salmon farming. Report APG/Cassell. under EU FAIR (ELSA) Project Organic Salmon Pro- Coull, J.R. (1972) The Fisheries of Europe: An Economic duction and Consumption: Ethics, Consumer Percep- Geography. London: Bell. tions and Regulation. European Union Commission, Cunningham, S., Dunn, M.R. and Whitmarsh, D.J. (1985) DG XII, November (unpublished report). Fisheries Economics: An Introduction. London: Ahoven, A. (1998) Unofficial fish supply in Finland. In: Mansell St Martin’s. A. Eide and T. Vaasdal (eds) Proceedings of the 9th Cutting, C.L. (1955) Fish Saving: A History of Fish Pro- Biennial Conference of International Institute of cessing from Ancient to Modern Times. London: Fisheries Economics and Trade, Tromso, Norway, Leonard Hill. July, pp. 737–9. International Institute of Fisheries DeChernatony, L. and McDonald, M.H.B. (1992) Economics and Trade. Creating Powerful Brands. Oxford: Butterworth Aitken, A., Mackie, I.M., Merritt, J.H. and Windsor, M.L. Heinemann. (1982) Fish Handling and Processing. Edinburgh: Donath, H., Wessells, C.R., Johnston, R.J. and Asche, F. HMSO. (2000) Consumer preferences for ecolabelled seafood in Alba, C. and Dennis, G. (2001) Scotland’s BSE? The Sun- the United States and Norway: a comparison. In: Micro day Times, 7th January, p. 11. Behaviour – Macro Results, 10th Biennial Conference Asche, F. and Sebulonsen, T. (1998) Salmon prices in of International Institute of Fisheries Economics and France and the UK: does origin or market place Trade, Corvallis, Oregon, USA, July. International matter? Aquaculture Economics and Marketing, 2, Institute of Fisheries Economics and Trade. 21–30. Ehira, S. and Uchiyama, H. (1986) Determination of fish Belknap, B. (2000) New research on the US salmon mar- freshness using the K value and comments on some ket. In: Micro Behaviour – Macro Results: 10th Bien- other biochemical changes in relation to freshness. In: nial Conference of International Institute of Fisheries D.E. Kramer and J. Liston (eds) Seafood Quality Deter- Economics and Trade, Corvallis, Oregon, USA, July. mination. Amsterdam: Elsevier, pp. 185–207. International Institute of Fisheries Economics and Euromonitor (1995) European Markets – Seafood and Trade. Fish. London: Market Research Europe. 58 Chapter 3

Fairlie, S., Hagler, M. and O’Riordan, B. (1995) The poli- Kinnucan, H. and Myrland, O. (1998) Optimal advertis- tics of overfishing. The Ecologist 25, 46–73. ing levies with application to the Norway-EU salmon FAO (1995) Code of Conduct for Responsible Fisheries. agreement. In: A. Eide and T. Vaasdal (eds) Proceedings FAO Fisheries Department. Rome: FAO. of the 9th Biennial Conference of International Insti- FAO (1997) Recent Trends in Aquaculture Production, tute of Fisheries Economics and Trade, Tromso, 1984–95. Rome: FAO. Norway, July, pp. 701–10. International Institute of FAO (1999) Yearbook of Fishery Statistics. Rome: FAO Fisheries Economics and Trade. and (www.fao.org/fi/). Kinnucan, H. and Myrland, O. (2000) What are the results FAO (2000) The State of World Fisheries and Aquacul- of the pan-European salmon marketing campaigns? ture. Rome: FAO and (www.fao.org/fi/). Presented to Seafood Opportunities in the New Mille- Fuller, D.A. (1999) Sustainable Marketing: Managerial- nium Aquavision Conference, Stavanger, Norway, Ecological Issues. London: Sage. May (unpublished). Gentles, P. and Roddy, G. (1999) Aquaculture and the Kumar, V. (2000) International Marketing Research. consumer. Presented to Aquaculture Europe Confer- New Jersey: Prentice Hall. ence, Trondheim, Norway, August. European Aqua- Lemon, D.W. and Regier, L.W. (1977) Holding of Atlantic culture Society. mackerel (Scomber scombrus) in refrigerated sea- Gentles, P. and Skeldon, S. (1994) Consumer trends in water. Journal of the Fisheries Research Board of European fish consumption. Paper presented to the Canada 34, 439–43. Fish ’94 International Aquaculture and Marketing Liabo, L. (1999) Aquaculture supply: focus on Atlantic Seminar, Bremen, June. Bremen: MGH Bremen GmbH. salmon. Paper presented to Salmon Summit ’99. Gordon, J.D.M. and Hunter, J.E. (1994) Study of Deep- Copenhagen, Denmark, September. Eastfish. water Fish Stocks to the West of Scotland. Oban: Lloyd, M. (1999) Salmon: the fish with a future – con- Scottish Association for Marine Science. sumer trends. Paper presented to Salmon Summit ’99. Guardian (2000) Jaws author campaigns to save the Copenhagen, Denmark, September. Eastfish. shark. 7th July, p. 11. Lockwood, S.J. (1984) The future of the mackerel. Gujja, B. and Finger-Stich, A. (1996) What price prawn? Buckland Lecture, University of Hull, 1 November. Shrimp aquaculture’s impact in Asia. Environment 38, Lockwood, S.J. (1988) The Mackerel – its Biology, 12–15. Assessment and the Management of a Fishery. Oxford: Harman, J. (2000) Is there reward for quality? Fish Trader, Fishing News Books. August, www.seafoodscotland.org/news.asp. Mackay, D.W. (1999) Perspectives on the Environmental Holden, M. (1994) The Common Fisheries Policy: Origin, Effects of Aquaculture. Presented to Aquaculture Evaluation and Future. Oxford: Fishing News Europe Conference, Trondheim, Norway, August. Books. European Aquaculture Society. Jaffry, S., Pickering, H., Wattage, P., Whitmarsh, D.J., MacSween, I. (1984) The interaction between fisheries Frere, J., Roth, E. and Nielsen, M. (2000) Consumer management and the marketing of fish. In: Expert Con- choice for quality and sustainability in seafood pro- sultation on the Regulation of Fishing Effort (Fishing ducts: empirical findings from United Kingdom. Mortality). FAO Report FIPP/R289. Rome: FAO, pp. Presented to Micro Behaviour – Macro Results, 9–19. 10th Biennial Conference of International Institute of McDonagh, P. and Prothero, A. (1997) Green Manage- Fisheries Economics and Trade, Corvallis, Oregon, ment: A Reader. London: The Dryden Press. USA, July. International Institute of Fisheries Econo- McGoldrick, P.J. (1990) Retail Marketing. London: mics and Trade. McGraw-Hill. Johnston, P. (2001) Quotas are impossible – Pos warn of Maddock, S. and Young, J.A. (1995) Catering for the Fish the return of black fish. Fishing News, 5 January, p. 3. Consumer – A UK Perspective. British Food Journal Kaylor, J.D. and Learson, R.L. (1990) Pelagic fish. In: R.E. 97, 21–5. Martin and G.J. Flicj (eds) The Seafood Industry. New MAFF (1999) Sea Fisheries Statistical Tables. London: York: Van Nostrand Reinhold, pp. 67–76. HMSO. Kent, R. (1999) Marketing Research: Measurement, Malhotra, N.K. and Birks, D. (2000) Marketing Research: Method and Application. London: Thomson Inter- An Applied Approach. Harlow: Prentice-Hall. national Business Press. Marshall, D.W. (1996) Food Choice and the Consumer. Key Note (1994) Key Note Market Review: UK Catering Glasgow: Blackie Academic Professional. Market. London: Key Note Publications. Marshall, D.W. and Currall, J. (1992) Consumer attitudes Key Note (2000) Fish and Fish Products. London: Key towards pelagic fish. In: J.R. Burt, R. Hardy and Note Publications. K.J. Whittle, Pelagic Fish – the Resource and its Marketing Fish 59

Exploitation. Farnham: Fishing News Books, pp. Ritzer, G. (1993) The MacDonaldization of Society. 260–6. London: Sage. Milgrom, P. (1989) Auctions and bidding: a primer. Schultz, D.E. and Kitchen, P.J. (2000) Communicating Journal of Economic Perspectives 3, 3–22. Globally – An Integrated Marketing Approach. Millar, S. (2001) Salmon farmers braced for clampdown London: Macmillan Business Press. on toxins. The Observer, 7 January, p. 1. Seafood Business (1998) Counter attack, 17, 1 and 22–8. Muir, J.F., Brugere, C., Stewart, J.A. and Young, J.A. (1999) Seafood International (1997) Health and wealth from the The solution to pollution? The value and limitations sea, 12, 9. of environmental economics in guiding aquaculture SFIA (2000) Key Indicators. Edinburgh: Sea Fish Industry development Aquaculture Economics and Manage- Authority. ment 3, 43–57. Shimp, T.A. (2000) Advertising Promotion: Supplemen- Muir, J.F. and Young J.A. (1998) Aquaculture and marine tal Aspects of Integrated Marketing Communica- Fisheries: will capture fisheries remain competitive? tions. Fort Worth, TX: Dryden. Journal of Northwest Atlantic Fishery Science 23, Smith, A.P., Young, J.A. and Gibson, J. (1999) How now, 157–74. mad cow? Consumer confidence and source credibility Nagle, T.T. and Holden, A. (1995) The Strategy and during the 1996 BSE scare. European Journal of Tactics of Pricing: A Guide to Profitable Decision Marketing 33, 1107–22. Making. New Jersey: Prentice-Hall. Strachan, N.J.C., Nesvadba, P. and Allen, A.R. (1990) Fish Nakamoto, A. (2000) The Japanese Seafood Market. SNF species recognition by shape analysis of images. Report No. 41/00. Norway: Bergen. Pattern Recognition 23 (5), 539–44. Naylor, R., Goldburg, J., Primavera Kautsky, N., Tacon, A.G.J. (1994) Dependence of intensive aquacul- Beveridge, M.C., Clay, J., Folke, C., Lubchenco, J., ture systems on fishmeal and other fishery resources. Mooney, H. and Mooney, H. (2000) Effect of aquacul- FAO Aquaculture Newsletter, April, No. 6, Rome: ture on world fish supplies. Nature 405, 1017–23, 29 FAO, pp. 10–16. June. Tapp, A. (1998) Principles of Direct and Database Mar- Norberg, H.M. (1996) Quality assurance schemes for keting. London: Financial Times Pitman. seafood – structure and implementation. Presented to UNDP (United Nations Development Programme) 8th Biennial Conference of International Institute (1997) Human Development Report 1997. New York: of Fisheries Economics and Trade, Casablanca, UNDP. Morocco, July. International Institute of Fisheries Walters, D. and Hanrahan, J. (2000) Retail Strategy – Economics and Trade. Planning and Control. London: Macmillan Business OECD (Organisation for Economic Co-operation and Press. Development) (1991) Environmental Labelling in Wandel, M. and Bugge, A. (1997) Environmental concern OECD Countries. Paris: OECD. in consumer evaluation of food quality. Food Quality Paquotte, P. (1998) New species in Mediterranean aqua- and Preference 8, 19–26. culture: is it an answer to the market demand for differ- Warde, A. and Martens, L. (2000) Eating Out: Social entiated products? Paper presented at the Societa Differentiation, Consumption and Pleasure. Italiana per il Progresso della Zootechnica, 33rd Inter- Cambridge: Cambridge University Press. national Symposium, New Species for Mediterranean Wessells, C.R. (1998) Assessment of the effectiveness Aquaculture, Alghero, Sardinia, April. of ecolabeling as a market based approach to Peacey, J. (2000) The Marine Stewardship Council Fish- sustainable seafood production. In: A. Eide and T. eries Certification Program: progress and challenges. Vaasdal (eds) Proceedings of the 9th Biennial Presented to Micro Behaviour – Macro Results, 10th Conference of International Institute of Fisheries Biennial Conference of International Institute of Economics and Trade, Tromso, Norway, July, pp. Fisheries Economics and Trade, Corvallis, Oregon, 475–83. USA, July. Wessells, C.R. and Anderson, J.G. (1995) Consumer will- Perez-Villareal, B. and Pozo, R. (1992) Fishing gear influ- ingness to pay for seafood safety assurances. Journal of ence on the spoilage of albacore (Thunnus alaunga) in Consumer Affairs 29, 85–108. ice. In: J.R. Burt, R. Hardy, and K.J. Whittle, Pelagic West, C. (1999) Marketing Research. London: Macmillan Fish – The Resource and its Exploitation. Farnham: Business Masters. Fishing News Books, pp. 170–77. Westlund, L. (1995) Apparent Historical consumption Reedy, J., Schullo, S. and Zimmerman, K. (2000) and future demand for fish and fishery products – Electronic Marketing: Integrating Electronic Resources exploratory calculations. Paper presented to Inter- into the Marketing Process. London: The Dryden Press. national Conference on Sustainable Contribution of 60 Chapter 3

Fisheries to Food Security. Kyoto, Japan, December. Mariojouls, C., Muir, J.F., Prothero, A., Reisch, L.A., Government of Japan & FAO. Smith, A.P. and Tveteras, R. (2000) The European con- Whitmarsh, D.J. and Young, J.A. (1985) Management of sumers’ understanding and perceptions of organic the UK mackerel fisheries. Marine Policy 9, 220– salmon production. Presented to Micro Behaviour – 36. Macro Results, 10th Biennial Conference of Interna- Whittle, K.J. and Wood, C.D. (1992) Utilisation of pelagic tional Institute of Fisheries Economics and Trade, fish for human consumption. In: J.R. Burt, R. Hardy Corvalis, Oregon, USA, July. International Institute of and K.J. Whittle, Pelagic Fish – The Resource and its Fisheries Economics and Trade. Exploitation. Farnham: Fishing News Books, pp. Young, J.A., Burt, S.L. and Muir, J.F., (1993) Study of the 238–53. Marketing of Fisheries and Aquaculture. Products Wilson, J.A. (1980) Adaptation to uncertainty and small in the European Community. Brussels: EC DG XIV. numbers exchange: the new England fresh fish market. Unpublished report to EU. Bell Journal of Economics 11, 491–504. Young, J.A., Crean, K. and Palfreman, D.A. (1992) Some Wise, M. (1984) The Common Fisheries Policy of the factors affecting the utilisation of small pelagics. In: European Community. London: Methuen. J.R. Burt, R. Hardy and K.J. Whittle, Pelagic Fish – The Young, J.A., Aarset, B., Beckmann, S., Bigne, J., Beveridge, Resource and its Exploitation. Farnham: Fishing M., Bjorndal, T., Bunting, M.J., McDonagh, P., News Books, pp. 254–9. 4 A History of Fisheries and their Science and Management

TIM D. SMITH

4.1 THE NATURE catch anything for breakfast?’ ‘No,’ they OF FISHING answered.

St Peter’s fishing trip on the Sea of Galilee, Fish are marketed in many ways, including infor- recorded by St John, illustrates many aspects of mal systems where potential purchasers ask after the nature of fishing, as Sahrhage and Lundbeck boats even before they come to the shore, as St suggested in their A History of Fishing (1992). Peter might have assumed was happening that Peterson’s (1993) contemporary translation of the morning. Gospel of St John goes like this: He said, ‘Throw the net off the right side of Simon Peter announced, ‘I’m going fishing.’ the boat and see what happens.’ They did The rest of them replied, ‘We’re going with what he said. All of a sudden there were so you.’ They went out and got in the boat. many fish in it, they weren’t strong enough to pull it in. Fishing requires capital and labour; someone owns the boat and the nets (perhaps St Peter), and others Where and when to go fishing has been asked since help with the work on the boats. In this case the people began fishing. In this story, the boat owner, total crew apparently numbered seven. Fishing is St Peter, chose when and where, at least until the frequently a family business; two of those helping trip was nearly over. Then Jesus directed the last St Peter that night were themselves sons of a net cast. His direction was quite precise, namely fisherman, Zebedee. off the right side of the boat. Jesus’ advice of where and when to fish has been interpreted as divine They caught nothing that night. intervention and alternatively explained as his having made use of an often-reported sudden Fishing is unpredictable; sometimes catches are appearance of fish when the prevailing westerly large, other times nonexistent, as in this case. wind suddenly ceases. The difference in catches Sometimes the fish are large, other times small. when the right and wrong decisions are made can Sometimes fishing will be poor for days or years, be enormous. For St Peter, Jesus’ direction resulted and is almost always seasonal. in a net that was too full for the men to load into the boat. When the sun came up, Jesus stood on the beach, but they didn’t recognize him. Jesus Then the disciple Jesus loved said to Peter, spoke to them, ‘Good morning! Did you ‘It’s the Master!’ When Simon Peter realized 62 Chapter 4

that it was the Master, he threw on some mining fishing strategies (Evans and Grainger, clothes, for he was stripped for work, and Chapter 5, this volume). They can also be used to dove into the sea. understand the nature of the fishing operations. For example, from St John’s story, we can obtain The motivations of fishermen for what they do, some insight into the fishing gear used. The maxi- and when they do it, are not simple. Here, St Peter mum size of the several cichlid species known as impetuously abandons his boat and crew, as well as St Peter’s fish is 3.5kg. Thus, the catch of 153 could the unexpected catch, showing that rational eco- have weighed as much as half a metric ton, appar- nomics of fishing are not always what motivates a ently near the upper limit of the fishing gear. Fur- fisherman. ther, a boat from that period recently recovered from Lake Kinneret near Ginnosar measured 8.2m The other disciples came in by boat for they long and 2.4m at its widest, consistent with the weren’t far from land, a hundred yards or so, likely crew size of seven and with the crew having pulling along the net full of fish. difficulty lifting that weight of fish into the boat. St John tells this story from the perspective of a Landing the catch is central to fishing and must be few fishermen in a single boat on one night’s fish- completed in a timely manner. Facilities for land- ing trip, but even in so limited a framework many ing range from beaches, as in this story, to ports aspects of fishing emerge: fishing gear, processing geared to efficient processing and marketing of the and markets, variability in catches, direction of fish (Young and Muir, Chapter 3, this volume). fishing, and fisheries statistics. If the story had This trip took place near the port town of been taken from a slightly wider perspective, say Ginnosar, on the southwest shore of the lake, that of the residents of Ginnosar, other aspects of near an area of warm salty springs that flow into fisheries would emerge. For example, St Peter’s the Sea of Galilee, now known as Lake Kinneret. large catch, especially if continued, might have been resented by other fishermen as decreasing the When they got out of the boat, they saw a fire fish available to them, or as flooding the market laid, with fish and bread cooking on it. Jesus and possibly depressing prices. Similarly, if St said, ‘Bring some of the fish you’ve just Peter’s fishing method was divine direction, caught.’ others might have seen it as unfair competition and argued to restrict its use. However, if Jesus’ Fish are marketed and used in many ways. While St method was merely keen environmental obser- Peter’s were immediately cooked for breakfast, vation, adoption of it might have begun an escala- they were more commonly salted or preserved tion of fishing intensity. in garum, a fermented sauce made of Spanish Here I briefly sketch the historical development mackerel or tuna guts. of fishing as a basis for the more complete descrip- tions in other chapters of many of the aspects of Simon Peter joined them and pulled the net fishing revealed in St Peter’s fishing trip. I focus to shore – 153 big fish! And even with all on the specific issues raised in this story, as well those fish, the net didn’t rip. as broader issues such as the economics of fishing, fishery management, and environmental Fishery statistics are widely reported, and usually and fishing effects on the supply of fish. I rely include the size and numbers of fish, as here. Often on, without citing repeatedly, two books: Sahrhage the location and the amount of fishing are rec- and Lundbeck’s A History of Fishing (1992) and orded, in this case one net for one night. Fishery my own Scaling Fisheries (1994). I have cited what statistics are used for various economic purposes, I see as their most important original sources, as for example for determining taxes and tithes well as additional sources not cited in those two (Holm 1996) and by fishing companies for deter- books. History of Fisheries 63

4.2 ORIGINS OF FISHING 4.2.1 Evolution of fishing methods Indications of fishing have been found in archeo- An enormous range of fishing methods was in use logical sites as early as the Late Palaeolithic period, by the time St Peter went fishing on Lake Kinneret. some 50000 years ago, revealing a long history of In addition to various forms of fish hooks, perhaps the use of fish by humans. More recently, fish and the first and most obvious form of fishing gear – a fishing have been depicted in rock carvings in myriad of nets, baskets, traps, spears, poisons and southern Africa and southern Europe dating from harpoons – had been developed. Most types of gear 25000 years ago. Based on this, Sahrhage and have been developed repeatedly, with the specific Lundbeck noted that fishing is one of the oldest form tailored to the peculiarities of species and professions, along with hunting. Unlike hunting, habitat. While the diversity of fishing gear is inter- however, fishing continued to be an important esting in itself (e.g. Misund et al., Chapter 2, this occupation even to modern times, and fishing volume), more astounding is the continual innova- methods have been repeatedly improved over the tion in form and efficiency. For example, even the millennia. basic fish hook evolved from early times into com- The development of fishing on all continents pound hooks, and into barbed hooks. Hooks were and in most cultures can be more clearly traced used singly and later in groups or gangs, with and since the early Mesolithic period, 10000 bc, and without bait, always being adapted and improved. can be seen in archaeological artefacts such as Similarly, the simplest beach seines and cast kitchen middens, paintings and fishing gear. For nets in use by St Peter evolved into complex purse example, in the port city of Ginnosar on Lake seines and trawls. A key element in the evolution Kinneret, where St Peter lived, excavations have of fishing gear was the vessels used. While a form of revealed consistent occupation and evidence of pair trawling (see Misund et al., Chapter 2, this fishing since the Bronze Age. By 4000 bc in some volume) is illustrated on the walls of an Egyptian areas the archaeological record is complete enough tomb from 2000 bc, the development of sail- to reveal the evolution of fishing gear. For example, powered vessels allowed this simple gear to prolif- the evolution from simple to compound fish hooks erate into many and increasingly complex forms has been demonstrated for the cultures on Lake (Anon 1921). This historical tendency to expan- Baikal in this period. sion was described succinctly in this century by In Asia the importance of fishing can be traced the oceanographer Henry Bigelow (1931): back for thousands of years, especially in the Yellow River in China and in the Inland Sea [f]or as [human] population multiplies in the off Japan. Fishing in China was primarily in countries bordering on those seas, fisheries fresh water, and tended towards the development will correspondingly advance in efficiency of of fish farming rather than catching technology. method, and an intensity of effort, extending However, illustrations of Chinese fishing methods at the same time farther and farther afield to showed them frequently to be unique: for example regions where the supply has hardly been multiple lines of hooks in complex arrays and tapped as yet. trained cormorants with neck rings. Chinese methods were subsequently used in Japan, initi- When Bigelow wrote he was looking back on a ally in the Sea of Japan. In the Americas fishing turning point in the development of fishing – the appears to have been imported with the colonizing rapid mechanization of fisheries initiated by the peoples, with rock paintings of fish and fishing as adoption of steam power in the late 19th century. early as 10000 bc in Patagonia. In the northern As the number of trawling and long lining fishing regions sealing and whaling developed, along vessels on the east coast of England decreased by with more traditional fishing, at least as early as 13% in the 10 years after 1889, the composition 2000 bc. shifted from 9% to 42% steam-powered vessels. 64 Chapter 4

Several changes in fishing gear followed rapidly. Most of the new steam-powered vessels were rigged for trawling rather than for long lining. At the same time, the greater and more consistent pulling power of the steam-powered vessels allowed the use of the new larger otter trawls to replace the smaller beam trawls. The new otter trawls were held open by water pressure on large in 1900 in 1875 flat ‘doors’ as the vessel moved forward, rather than by a fixed head-beam (Misund et al., Chapter 2, this volume). These doors were designed after in 1865 devices used in Irish and Scottish salmon (Salmo salar) trolling gear, and were tied to the cables be- tween the ship and the net. Introduced in Scotland in 1892, this markedly more efficient trawl spread in 1845 rapidly as fishermen continued to compete against one another for higher and higher catches. Area As Bigelow pointed out, not only did fishing trawled vessels and gear improve over time, but fishing in 1833 tended to expand into regions ‘farther and farther afield’. Even before the east coast English fleet ex- panded in size and power in the late 19th century, Fig. 4.1 Alward’s map of the North Sea showing (cross- the areas fished had also been expanding (Fig. 4.1). hatch patterns) the years in the 19th century when The spatial expansions in the later periods were English trawlers first began fishing in increasingly more allowed by the increased capability of the larger northerly areas. (Source: from Wimpenny 1953.) steam-powered vessels, but even before that tech- nology, expansion was evident. Such expansion has been seen in many fisheries, frequently driven by declining catch rates as the fish populations in ing vessels and gear are described by Misund et al. the nearer areas are reduced. Choice of where to (Chapter 2, this volume). fish then frequently becomes a trade-off between travel costs to and from suitable ports and markets 4.2.2 Evolution of ports and catch rates in different areas. This trade-off has and marketing often pushed vessels ever further into the sea, as seen in the English fleet in the 19th century. St Peter could have expected to sell his catch by In the 20th century there were many additional landing on the beach, and this is still often the case advances in fishing vessels and gear. In addition to in many artisanal fisheries. However, villages and a gradual shift from steam to diesel engines, some towns often developed in favourable locations for vessels began to process the catch on board. In the landing and marketing fish. For example, the 1950s hydraulic winches and the ‘power blocks’ development of favourable transportation systems began to be used, greatly increasing the ability of can create or ensure the viability of ports, as oc- fishermen to deploy the gear and to lift the filled curred in 1860 in the English town of Grimbsy nets back aboard. Similarly, after the Second World when the rail lines to London were established. War echo location equipment became increasingly Further, the expansion of fishing grounds some- available, allowing fishermen to target their prey times creates new ports. before setting their gear, rather than searching The existence of transportation systems and with the gear. The results of this evolution of fish- markets also has spurred the development not just History of Fisheries 65 of particular ports, but of entire regional fisheries. to fish in the USA in the 1930s by Clarence Birdseye, For example, the North Pacific halibut (Hippoglos- and along with marketing New England haddock sus stenolepis) fishery began to expand in (Melanogrammus aeglefinus) as fillets rather than Vancouver and Seattle after the 1880s, with the in the round, contributed to the expansion of the completion of transcontinental railways in market for this formerly disregarded species. North America, but only as the North Atlantic halibut (Hippoglossus hippoglossus) fishery began to collapse (Goode and Collins 1887). 4.3 OVERFISHING While the existence of a natural harbour is usu- ally a prerequisite, that alone is not always enough. One effect of the rapid increase in fishing power Sometimes fisheries are encouraged as a part of towards the end of the 19th century was increasing social policy. For example, in the late 18th century arguments about the possibility of too much fish- a benevolent British Fisheries Society in London ing, or, using John Cleghorn’s 1854 term, overfish- raised funds and established the Scottish port of ing. Although a suitable definition of this word Ullapool in northwest Scotland in a successful proved elusive, its possibility was nonetheless attempt to provide employment opportunities in hotly debated for decades after Cleghorn, who, to that region (Dunlop 1978). the dismay of his neighbours, argued that herring The processing, preservation and marketing of near his village was overfished. Often the debate fish have evolved along with fishing gear and ves- was inspired by conflicts between fishermen using sels. While fish was undoubtedly first consumed different ports or gear. A more efficient net invent- fresh, much ingenuity has been applied to develop- ed in 1872 by an enterprising sardine fisherman in ing methods of processing. Salting fish to dry and Douarnez, France, for example, was blamed by preserve it has been done since time immemorial, other fishermen for being too efficient, an efficien- contributing, for example, to the establishment cy thought to cause subsequent declines in catches of rather unpleasant salt mines throughout the due to overfishing. The arguments were sufficient- Roman Empire. Historically many other methods ly vociferous that the new design was abandoned. have also been used, such as fermentation and As fishing efficiency increased towards the end packing fish in oil. The specific methods of pre- of the 19th century, the overfishing argument serving imparted their own sometimes peculiar moved into the scientific community. It was de- flavour. For example, the garum used when St bated, for example, in 1883 during the Great Inter- Peter was fishing was produced in factories in Italy, national Exhibition in London. The disputants Spain, North Africa and Turkey, and was shipped were two of the leading zoologists of the time, in amphorae through the Mediterranean region. Thomas Huxley and Ray Lankester. Huxley had The amphorae were labelled with the manufac- long argued for reducing the seemingly arbitrary turer’s name, as well as the date and the quality. regulations on fisheries in England, complaining Some forms of garum were particularly highly that there was no scientific basis for them and that valued, and the method continued to be used until they unnecessarily hurt fishermen. In the debate, the 15th century in some areas. he argued that the great fecundity of fishes implied Canning was developed early in the 19th cen- that ‘probably all the great sea fisheries are inex- tury in response to a prize offered by the French haustible’ (Huxley 1884). Lankester was also Revolutionary Directory for a new method to pre- impressed with the fecundity of fishes, but saw serve food. Experiments with fish in the United the production of young fish as important in itself: States and Britain resulted in the first canned com- ‘the thousands of apparently superfluous young mercial fish product around 1822. The technique produced by fishes are not really superfluous, but was combined with the technique of packing fish have a perfectly definite place in the complex in oil and was used for French sardines (Sardina interactions of the living beings within their pilchardus), beginning in 1825. Freezing was applied area’ (Lankester 1884). 66 Chapter 4

Although the Huxley–Lankester debate settled 1905). Although this was one of the few ‘effective nothing, it did give impetus to the already develop- organizations’ in the English government at ing scientific investigations of fisheries. By the that time, scientists examining these new data middle of the 19th century concerns about fluc- were not impressed. Only total catches of several tuations in fish catches had stimulated the ap- species were recorded, and those in a myriad of ill- plication of science to fisheries. Russian and defined market size classes. Although many im- Norwegian studies of the fisheries and the basic provements in the collection of fishery statistics life history of fishes began in the 1850s. Investiga- were subsequently made, adequate statistics for tions within the US Fish Commission began scientific purposes have proven difficult and in the 1870s and initially took a broad ecological expensive to collect, and too frequently have been perspective, outlined by the Commissioner, unrepresentative of the continually changing fish- Spencer Baird (1872). The initial scientific eries. Difficulties in interpreting data collected in problem was an almost complete lack of data. the ports led to placing technicians aboard fishing The research programme initiated in 1885 by vessels to record more detailed data, an even more the reorganized Fisheries Board of Scotland, for expensive approach (see Evans and Grainger, example, was designed to remedy this, and in- Chapter 5, this volume, for more detail of data cluded both the collection of fishery statistics collection). and (interestingly at the insistence of Huxley) an experimental trawl survey in areas open and 4.3.2 ‘Impoverishment of the seas’ closed to fishing to determine the effects of fishing. The Huxley–Lankester debate was resolved only after the data from the Scottish fishing experiment 4.3.1 Measuring Fishing were analysed, and reanalysed, over the waning years of the 19th century by three scientists, St Peter’s fishing trip can be described in many Thomas Fulton, William Mclntosh and Walter ways: for instance, St John recorded the amount of Garstang. Fulton’s initial analysis was immedi- time spent fishing, the number of crew members, ately attacked by Mclntosh, who argued that the and the size and number of fish caught. Other mea- experimental data were worthless to begin with sures that have proven useful in fisheries include because of how they were collected (Mclntosh the size of the nets used and the number of times 1899). Further, despite his initial support, he later the net was cast and retrieved. argued that the experiment had been unnecessary Such measures have been used to set taxes, for because it was apparent on inspection of the fecun- example on the basis of the number of oars owned dity of fish that Huxley had been correct. This by a fisherman, or to determine the tithes due the point of view, however, too often became entan- local parish (Holm 1996). They have also been used gled with vested interests, who were against the as a basis for settling international disputes for regulation of fishing and relied on extensive appeal determining investment strategies, and for propa- to authority rather than to scientific demon- ganda to promote investment and colonization stration: McIntosh apparently convinced few (Innis 1940). While statistics have been recorded scientists. by governments and companies for specific In the middle of the argument between Fulton purposes since time immemorial, the Huxley– and McIntosh, Garstang was hired to continue a Lankester debate focused attention on the need for fisheries research programme at Grimbsy. He un- better fishery statistics to underlie development of dertook to reanalyse the Scottish data because he public policy. felt that Fulton’s analysis had been too superficial, In the 1880s the English Board of Trade began especially in terms of stratifying by season. He also recording additional fishery statistics (Johnstone began an analysis of the Board of Trade’s fishery History of Fisheries 67 statistics, which had now been collected for a the discussions leading to ICES, the marriage decade. between the two legs proved to be weak at best In 1900 Garstang reported his conclusions from (Chapman 1962). these two lines of analysis in a paper titled ‘The The ICES Biological Programme was subse- Impoverishment of the Sea’. The Scottish experi- quently elaborated to include studying fish growth ment, he argued, directly contradicted Huxley. and maturity, migration pattens, abundance and Not only could fishing reduce the abundance of reproduction, but with only limited reference fishes, but the experimental data also suggested, in to the Hydrography Programme. The biologists support of Lankester’s ‘complex interactions’, that began by trying to define overfishing, noting that fishing one species could indirectly result in the in- this could mean either that fish were not allowed crease of another. Garstang also showed that not to grow to an efficient size, or that too many were only could fishing affect fish, but in the period caught to allow adequate reproduction (Petersen 1885 to 1895 fishing actually did decrease fish 1903). Further, they debated if overfishing should abundance. Using fishery statistics collected in be defined entirely on biological grounds, or if the port of Grimbsy, he also showed that the catch economics should enter in (Kyle 1905). In the end of fish per vessel (standardized for the changes in they adopted a simple definition for overfishing the fleet with the advent of steam power discussed as ‘too severe fishing,’ and particularly the situa- above) had declined by 33%, while the fishing tion that ‘more fish or better qualities of fish intensity had increased by 150%, further de- were taken away than the natural production monstrating the effect of fishing on the fish could replace’. populations. To provide scientific information to support in- ternational fisheries agreements, ICES fishery sci- 4.3.3 The Great Fishing Experiment entists began an ambitious research programme. They adapted and extended many of the field Garstang’s resolution of the Huxley–Lankester de- and statistical methods that had been developed bate did not resolve the question inherent in the in the 1890s, including marking and recapturing fact that ‘overfishing’ was still undefined. Pursuit fish, determining ages and reproductive condition, of a suitable definition became the task of the first identifying different populations or stocks of intergovernmental marine scientific body, the In- fish (Sinclair and Solemdal 1988), and conduct- ternational Council for the Exploration of the Sea ing research-vessel trawl surveys to determine (ICES) (see Rozwadowski 2002). Arising out of the abundance and plankton sampling surveys to spirit of internationalism of the 1890s, ICES was determine egg production. They compiled catch established in 1902 on two legs, hydrography and statistics, measured fish at sea and on the docks, biology. Hydrography was the major motivation and, of course, wrote reports. for the initial discussions that led to the formation Over the first pivotal decade, ICES scientists of ICES. The basis for the ICES Hydrography Pro- eventually focused on a single issue, the status of gramme had been well worked out by 1898, draw- the North Sea plaice (Pleuronectes platessa). By ing on joint work among Scandinavian scientists 1913 they were convinced that overfishing was in throughout the 1890s. In contrast, the Biology Pro- some sense occurring, concluding: ‘the fear of an gramme was initially less central, being added on already existing or impending overfishing seems, during the negotiations with carefully crafted lan- therefore, perfectly legitimate and justifiable. . . . guage to ensure that the work to be done by ICES [T]o deny this danger is to close the eyes intention- would help ‘promote and improve the fisheries ally’ (Heincke 1913). The scientists had trouble, through international agreements’. While the however, in agreeing how their proposed reduc- importance of using oceanography to support the tions might be made, deciding in the end that, as an biological programme’s goals was often noted in experiment, a minimum size limit might be set 68 Chapter 4 at 25 or 26cm, perhaps phased in over a period of 1907 years. 14 The 1913 recommendation was met with little enthusiasm among ICES countries, even with the 12 possibility of a gradual imposition of size limits. This lack of enthusiasm was not important, how- 10 ever, because the guns of August 1914 signalled the 1914 beginning of an altogether more drastic experi- 8 ment, the reduction of fishing activity during the 1902–7 1919 6 Great War. This experiment, the Great Fishing Percentage Experiment, was far more telling than that 4 proposed by ICES. Even as the war ended, scientists began assem- 2 bling fishery statistics. Anticipating difficulty in- terpreting the postwar market category statistics, 0 scientists collected some additional data by 10 15 20 25 30 35 40 45 50 55 60 65 putting technicians aboard some British fishing Length (cm) vessels to record the length of the fish caught. The analysis of the before and after data followed the Fig. 4.2 Borley’s graph of the percentage of plaice in each length category caught in the fishery on Dogger approach used by Garstang 20 years earlier, com- Bank in the period 1902–7, and in 1907, 1914 and 1919. paring the weight of fish landed in each market cat- The greater lengths of plaice caught in 1919 demonstrat- egory and the number of days fished. The results ed that the higher postwar catch rates were due to the were even more dramatic than Garstang’s, show- growth of fish during the war years, when fishing ing that the weight of catch per day fished in- intensity was low. (Source: redrawn from Borley et al. creased by 500% for the largest market category. 1923.) Further, the cause of the increase was also clear, with fish having become so much larger that the smallest fish landed after the war were larger than thermocline on the west side of the lake where St the modal size of fish landed before the war (Fig. Peter was fishing. This could explain the miracu- 4.2) (Borley et al. 1923). The result of the Great lous catch. Characteristic of oceanographic expla- Fisheries Experiment was clear: fish populations nations, however, this attractive explanation does can recover when fishing is reduced. not work very well because the westerly wind tends to abate in the afternoon, not in the morning 4.3.4 Operational fisheries when he was fishing. oceanography In Japan, hydrography was pursued to help locate where fish congregate and might more While ICES scientists over the first decade of the easily be caught. This interest arose after the 1901 20th century focused on overfishing of North Sea ICES preparatory meeting in Oslo, where Japan plaice, other scientists focused on an issue that was represented by an observer. By 1909 Tasaku arose in St Peter’s fishing trip: predicting when and Kitawara had expanded Japan’s nascent fisheries re- where to fish. In Lake Kinneret, the sudden appear- search programme to include systematic oceano- ance of fish associated with the cessation of the graphic observations. By 1918 he had identified a prevailing winds is a good example. Limnological pattern in fish distribution called Kitawara’s Law, studies have since shown that when the prevailing namely that fish tend to accumulate around westerly wind blowing across the north–south ori- oceanic fronts (Uda 1962, 1972). Kitawara’s Law entated lake suddenly stops, a seche is created that has stood the test of time, and represents one focus results in a predictable upward movement of the of fisheries hydrography, recently referred to as History of Fisheries 69

‘operational fisheries oceanography’ by Kendall and even decline. No matter what individual and Duker in their history of this field (1998). fishermen do then, they can no longer maintain their income, and economic difficulties set in (see Hannesson, Chapter 12, this volume). Fishermen with large mortgages on their vessels cannot fish 4.4 A BASIS FOR less, for then they would not be able to make their MANAGEMENT: GRAHAM’S payments to the bank. Banks become edgy because LAW OF FISHING the value of the vessels is tied to the abundance of the fish. Incomes of both vessel owners and crew In 1900 Garstang had shown that increased fishing decline. Calls for government intervention soon could reduce fish abundance and in 1920 ICES sci- develop. entists demonstrated that the reverse could also In the North Sea, the rapid postwar rebuilding occur – decreased fishing could allow fish abun- of the fishing fleets in the face of the scientists’ evi- dance to increase. This completed the loop, and dence demonstrated that the very thing that ICES many scientists involved in the 1920s had con- scientists had agreed not to address, the economics fidence that the obvious implications of this of fishing, resulted in complex motivations that demonstration would quickly be embraced by had not been adequately accounted for in the man- ICES countries. However, as St Peter’s fishing trip agement of this fishery. showed, the motivations of fishermen are complex and not always rational. Although greater catches 4.4.1 Early Management Institutions and higher catch rates had been demonstrated to be possible, there was no agreement after the first Although few, if any, marine fisheries were active- World War to limiting fishing in order to take ad- ly managed in the 19th century, fishermen vantage of the greater efficiency. Instead, there was nonetheless had to contend with a sometimes a rush to expand the national fleets rapidly, capital- complex suite of rules and regulations. In England, izing on the huge immediate profits to be made. many of these were swept away in the 1880s, being What caused this intentional closing of the judged not to be supported by science (Huxley eyes, a behaviour that ICES scientists had identi- 1884). Fisheries expanded rapidly in the early fied earlier when they attempted to give advice in 1900s in the North Sea, and it was the result of that 1913? One explanation is the pattern that Michael expansion that Graham had to draw on in formu- Graham would later describe as the Great Law of lating his Great Law of Fishing. He was also aware, Fishing (Graham 1943). In his work in the 1920s however, that his law also applied to other fish- and 1930s, Graham saw a recurring pattern, one eries. For example, it applied in whale fisheries, that seemed to play out in fishery after fishery. The where the successive exhaustion of sperm whaling underlying observation was what had been so grounds (Physeter macrocephalus) in the North clearly demonstrated by Garstang in 1900 and Atlantic was clear to Herman Melville. Melville in again after the Great Fishing Experiment: catches Moby Dick attributed declining catch rates to the do not continue to increase in proportion to the whales moving to new grounds to avoid the amount of fishing. Rather, eventually further in- whalers. As Johan Hjort later showed, however, creases in fishing result in smaller increases in the exhaustion of the North Atlantic whaling catches, and profits begin to decrease. To maintain grounds following the invention of the explosive their income, fishermen invest in bigger and better harpoon in the 1870s was caused by removals that nets and boats, increasing their fishing power and were too great, not by movement (Hjort 1933). The at least temporarily allowing an edge against other Antarctic Ocean fishery for blue (Balaenoptera fishermen. musculus) and fin (Balaenoptera physalus) whales Increased long enough, fishing intensity reaches began in earnest in 1925, following the develop- a level beyond which catches no longer increase, ment of the stern-ramped factory ships capable of 70 Chapter 4 processing harpooned whales at sea, which freed and the US before the First World War. A treaty was the whalers from their increasingly restrictive eventually signed in 1924, establishing the North shore-based factories. Management of this inter- Pacific Halibut Commission to manage the fish- national fishery was attempted through the ery, and allowing for the hiring of scientific staff. League of Nations, and the International Council The Commissioners immediately hired William for the Exploration of the Sea was drawn in to con- F. Thompson away from the California Fish and duct the needed scientific inquiry. However, man- Game Commission to oversee the research. Draw- agement was too weak and this fishery would also ing on the (by now standard) fishery research meth- follow Graham’s Law (Small 1971; Schevill 1974). ods of mark and recapture, age determination, The effects of Graham’s Law also began to be oceanographic and egg surveys to establish spawn- seen by the 1920s in the rapidly developing New ing area, fishery landings statistics, and analysis of England haddock fishery. Haddock had always fishermen’s deck logs, Thompson soon demon- been a minor component of the New England strated that the first part of Graham’s Law had fishery, which had focused on Atlantic mackerel already occurred. (Scomber scombrus) and cod (Gadus morhua). The evidence was in the fishery statistics after Frozen haddock fillets, however, had expanded the the three marked periods of increase in fishing ef- market and the otter trawl landings increased by fort beginning in 1895, 1910 and 1918. Following 150% over five years. At the same time, at-sea ob- each of these effort increases, the catches also servations on fishing vessels revealed that large increased. After the last two, however, the catch numbers of small fish were being discarded be- rates declined. Armed with this information, cause they were too small for the market. Esti- Thompson was repeatedly able to influence his mates of the discard rates were as high as 30% of Commissioners to expand the length of the ini- the landed catch. William Herrington, then work- tially agreed seasonal closures, thus successfully ing for the US Bureau of Commercial Fisheries, avoiding the second part of Graham’s Law. showed that after 1927 the catch rates began to de- The management institutional structures de- cline, as Graham’s Law predicted. The declines veloped in the first half of the 20th century varied prompted concern, the concern prompted re- greatly. In the North Sea, ICES provided manage- search, and the research prompted more concern. ment advice, usually along the lines of size limits, But, as in the North Sea, the concerns never be- and international agreements were forged, and re- came sufficient to prompt actual management of forged. In the US the individual state governments, the fishery. Instead, management measures were not the national Bureau of Fisheries, had authority discussed and evaluated. Further, experimental to regulate fisheries. In New England, the neces- trawling was conducted to measure the effect of sary international discussions with the several possibly increasing the size of mesh in the nets, in involved countries were limited to a formal the hopes that the discarding could be reduced. dialogue among the scientists. Similarly, the im- However, although the fishermen informally portant California sardine (Sardinops caerulea) agreed ‘to avoid, whenever possible, grounds fishery was regulated directly by the California where small fish predominate’, they also regularly Legislature, with the California Fish and Game resurrected Huxley’s long-disproved argument Commission having only limited authority on that fisheries were infinite. how the catch might be used. Graham’s Law of Fishing was not inevitable, however, and it was forestalled in at least one fish- 4.4.2 Methods of management ery, the US and Canadian North Pacific halibut fishery. The fishery expanded so rapidly after the Many methods of controlling fishing were being turn of the 19th century, with completion of the tried in the early 20th century. Even as early as rail access to the east coast, that negotiations for 1871, the US Fish Commissioner had proposed international regulations began between Canada weekend closures for the scup (Stenotomus History of Fisheries 71 chrysops) fishery in Massachusetts and Rhode focus of research throughout the first half of the Island (Baird 1872). In the failure of state gov- 20th century. In 1927 the head of the US Bureau of ernments to agree to restrictive management Fisheries defined the agency’s research objective approaches, the Commission began advocating not as implementing management, but as research the politically popular but unproven method of re- to ‘devise means and develop methods of effecting leasing newly hatched cod fish larvae to supple- real fishery conservation’. The values of the vari- ment natural production (Baird 1887; Dannevig ous forms were formally debated by Herrington 1910; Taylor 1999). Size limits had often been sug- (1943) and Nesbitt (1943) in the early 1940s, setting gested, ranging from the informal ‘biological limit’ the stage for post-Second World War developments. advocated in the North Sea in the 1890s with a The success of the North Pacific Halibut Com- minimum size equal to the size of sexual maturity, mission was closely watched by fishery biologists to the ever-fluctuating minimum marketable size. in Europe, and served as a model for management Attempts to obtain compliance with size limits after the Second World War. After the war, a series often forced discarding of too small fish, but of bilateral and especially multilateral fishery encouraged the use of larger mesh sizes, which agreements were negotiated. For example, the allowed smaller fish to escape unharmed. International Whaling Commission established Area and season closures were also tried, with control, albeit weak control, over especially the varying success. The North Pacific Halibut Com- Antarctic Ocean whaling. The International Com- mission was successful by continually increasing mission for the Northwest Atlantic Fisheries was the seasonal closures to the point of having a sea- established to manage trawl fishing off Greenland, son diminished to weeks and then days. Salmon Canada and the United States. The Inter-American fisheries have often been regulated by closing and Tropical Tuna Commission and the International opening river fishing in a manner designed to allow Commission for the Conservation of Atlantic a certain number of fish to escape to the streams Tunas were established to manage tuna fishing to spawn. By 1938 ICES scientists had developed in the eastern tropical Pacific and in the Atlantic, a management approach for Antarctic whaling respectively. based on limitations of fishing effort, although in the end only minimum size limits were agreed to. Limits on total catches were also proposed, for 4.5 SCIENTIFIC BASES example by the California Fish and Game Com- FOR MANAGEMENT mission for the sardine fishery. But political pressure on the California legislators from those As Director of Fisheries in Norway in the early profiting from that fishery was sufficient year after 1900s, Johan Hjort had undertaken to develop a life year to prevent any limits being set. insurance plan for fishermen. Based on that experi- Of these pre-Second World War attempts ence, in 1907 he made what he later termed a ‘bold at management, only a few were successful suggestion’ to his ICES colleagues, a suggestion (Engholm 1961). The bilateral North Pacific that was not well received but became a turning Halibut Commission was one, with its own group point for fisheries science (Hjort 1914). His pro- of scientists. The multilateral Alaska Fur Seal posal was that they apply the science of Vital Commission (Callorhinus ursinus) was another, Statistics to fish. although it relied on the various national scientists Hjort identified three types of information that rather than establishing a dedicated scientific are routinely studied in Vital Statistics or demog- group. But the complexities of the fishing patterns raphy: birth rate, age distribution and migration. and economics of North Sea nations prevented ef- Scientists had already developed methods of mea- fective management, as did the lack of federal con- suring the first, the fecundity of fish, the sheer trol in the USA. The ideal form for management enormity of which had fascinated both Huxley and controls was a constant topic of discussion and Lankester. They had also developed methods of de- 72 Chapter 4 termining fish migrations by marking the fish and 4 6 8 10 12 14 16 18 then seeing where they are subsequently caught. Hjort himself had even had some success in deter- 20 1907 mining the age distribution of a population based 10 on the modes in the length distribution. ICES scientists were sceptical, however, be- 20 cause the latter two components of Hjort’s pro- 1908 10 posal were not well known. Although egg produc- tion could be measured, it was apparent that most of those eggs did not survive. Determining how 10 1909 many did seemed an insurmountable problem. Second, although the age distribution could be 30 determined at least roughly from length dis- 20 1910 tributions for fish with seasonal spawning, as Percentage Hjort had shown, the determination of the age of 10 an individual fish also seemed an insurmountable problem. 10 1911 Undaunted, Hjort persisted and he and his col- leagues eventually showed that the key to both 10 1912 problems, determining survival and determining 10 age, was in the rings on scales or on otoliths (see 1913 Jobling, Chapter 5, Volume 1). By counting the 10 1914 rings they knew the ages, and by looking at the ages of herring over several years, they showed that the 4 6 8 10 12 14 16 18 fish of some ages or year classes were much more Age of spring herring (years) frequently seen than others. This suggested that in some years the numbers or survival of eggs must Fig. 4.3 Hjort’s graph of the percentages of spring herring aged 2 to 18 years in catches made from 1907 to have been much higher than in others. For exam- 1914. The regular advance of the peak percentages from ple, by 1908 the abundance of four-year-old spring- year to year and the relatively low percentages of other spawning herring, those spawned in 1904, was age classes demonstrated the persistence of the extraor- greater than any other age. Further, the 1904 year dinarily abundant herring that hatched in 1904. (Source: class continued to predominate, with the peak ad- from Hjort and Lea 1914.) vancing in age year after year (Fig. 4.3). From demo- graphic data like these, fisheries scientists rapidly confirmed Hjort’s suggestion that the variability some relevance to overfishing, which had become in catches, which prompted the application of sci- the driving focus of ICES. The causes of these fluc- ence to begin with, was due to extreme variability tuations were hypothesized to operate during the in survival of young fish (see Myers, Chapter 6, first early days and weeks after the young fish Volume 1). hatched. To test this hypothesis, biologists have compared measurements of currents, temperature 4.5.1 Recruitment fisheries and other oceanographic variables extracted from oceanography the oceanographer’s records, with year class strength measurements derived from fishery sta- Many patterns in year class strength have been tistics or from research surveys (see Myers, Chap- identified by fisheries scientists since Hjort’s con- ter 6, Volume 1). vincing demonstration of year class fluctuations. The application of oceanography to the problem The ICES Hydrography Programme started to have of the variability of year classes was pursued in the History of Fisheries 73

1920s in Northern Europe and North America. In His approach allowed the determination of the some cases this involved the fishery biologists effect of protecting younger fish on the total poten- doing a little oceanography on the side, while in tial catch over the lifespan of a year class of fish. Un- other cases major research programmes were de- fortunately, published on the eve of the Russian veloped. In the 1920s Henry Bigelow, working Revolution and only in Russian, Baranov’s analysis from Harvard University, pursued the oceano- was apparently not understood until about 1938. graphy of the Gulf of Maine for the US Bureau of His basic equation was built around equation Fisheries. Oscar Sette followed Bigelow with a (13.5) in Sparre and Hart (Chapter 13, this volume). Depression-interrupted study of the survival of The second time the joint effects of growth Atlantic mackerel off New York, and continued and mortality were worked out was by William this focus in a longer-term study of the California Thompson by the mid-1930s (Thompson and Bell sardine fishery beginning in the 1930s. The prob- 1934). In addition to showing that the catch rates lem of explaining the causes of the variability of had declined after effort had increased (as de- year class strength continued to fascinate biolo- scribed above), Thompson demonstrated the inter- gists and oceanographers throughout the 1930s action between growth and mortality numerically. and 1940s, but the answers were few and the appli- He showed that the catch rates had to decline be- cations to the practical questions of the ICES Bio- cause of the increased mortality. Baranov’s and logical Programme were limited. This continued Thompson and Bell’s analyses provided the start- to be the situation at least up to recent times, as ing point for the development of more sophisti- discussed by Myers (Chapter 6, Volume 1). cated models of the trade-off between the age of the fish caught and the ultimate yield. These de- 4.5.2 Modelling age structure velopments are reviewed by Shepherd and Pope (Chapter 8, this volume). The first development A more fruitful approach to determining the status was by a wartime colleague of Graham, H.R. of fisheries was to observe the consistent progres- Hume, who in response to questioning by Graham sion of strong year classes as members aged. This had completed a simple mathematical analysis of was especially clear in the case of the 1904 year the trade-off between catch from a year class and class of herring, which continued to support the fishing intensity. At Graham’s insistence, in 1947 fisheries for several years after Hjort published his Hulme and two students Graham had just recruit- results in 1914. As this year class aged, two things ed, Raymond Beverton and Sidney Holt, published happened. The first and obvious thing was that a seminal paper (Hulme, Beverton and Holt 1947) members of the year class were dying. Some died on the trade-off between catch and fishing rate (Fig. naturally in the course of things, while others were 4.4). Although growth was represented in an overly caught and marketed. The second thing was that simplistic manner, the graph showed succinctly the fish grew both in length and in weight. Implica- that yield from a year class must ultimately de- tions of these two processes for fishing were appar- cline as the fishing rate is increased, explaining ently worked out twice. The first was by Fëdor at least part of Graham’s Law of Fishing. Increasing Baranov in his 1918 paper, ‘On the question of the the fishing rate results first in increasing yield biological basis of fisheries’. Baranov observed that in weight, but increasing at a rate that is less increasing fishing would increase the total mor- than proportional of the fishing rate. This is seen in tality rate, and thereby decrease the proportion Fig. 4.4 by the yield curve beginning to bend over of a year class surviving to older ages. He also ac- as fishing effort increases. Then, as fishing rates counted for the growth of individuals, assuming increase further, the yield begins to decrease, as for convenience that fish increased steadily in seen by the decline of the right-hand side of the length, and elegantly showed how to compute the yield curve. The most that can be caught, of course, trade-off in potential catches due to the decreasing is at the fishing rate corresponding to the peak of numbers and increasing individual sizes with age. this curve. 74 Chapter 4

Volterra had worked on the physics of elasticity. One was Sir Ronald Ross, who received the Nobel Prize in 1902 for his studies of malaria. He later de- veloped a mathematical model of mosquitos, malar- ia and humans. Using his modelling approach, which he termed the a priori method, he showed that malaria could only survive when the mosquito

Yield in weight population was above a certain density, but that it could increase very rapidly with only small increas- es in mosquito density above that level. Similarly, entomologist William R. Thompson (unrelated to the William F. Thompson of halibut fame) developed 0 Fishing rate mathematical models for studying the effects of con- Fig. 4.4 Hulme’s graph showing the predicted increase trol methods on corn borers. and subsequent decrease in the fishery yield in weight A physical chemist, Alfred Lotka, applied (vertical axis) as the fishing rate (horizontal axis) mathematics to biology in a much more general increases, as implied by a simple mathematical model. way. He described his approach to mathematical This model demonstrated that fishing harder could modelling in his 1925 book Elements of Physical result in lower catches. (Source: from Hulme et al. 1947.) Biology. That book included as a special case exactly the problem that D’Ancona had posed to 4.5.3 Population dynamics Volterra, one that Volterra solved and published in 1926. Lotka was quick to point out in response to Even though sales in the Trieste fish market de- Volterra’s paper that not only had he already solved creased over the years of the First World War, that problem in his book, but in fact had shown Umberto D’Ancona, an enthusiastic young that the behaviour of the model becomes more Italian marine biologist, continued collecting his complicated with the addition of a second compet- fishery statistics. He noticed that as the total sales ing prey species harvested by the fishermen (for the declined the mix of species changed, with the pro- equation, see Pauly and Christensen, Chapter 10, portion of predatory fish increasing steadily. As this volume, equation 10.1). sales began to increase after the war, he also saw The mathematical modelling of Ross, W.R. that the proportion of predatory fish declined. Thompson, Lotka and Volterra had demonstrated These changes were reminiscent of Lankester’s that population problems were complex, as ‘complex interactions’. Baranov also suggested tak- Lankester had pointed out in 1883. Indeed, their ing another perspective on interpreting the effects sophisticated models have become fundamental of war on fisheries by explicitly (and unusually at to ecological studies (Kingsland 1985). However, it that time) including the fishing industry (Baranov was the much simpler approach of a demographer, 1926). Both D’Ancona’s and Baranov’s approaches Raymond Pearl, that was picked up by fishery were far more complex than the approach ICES biologists. Pearl’s wartime experience as Chief scientists like Graham had taken in interpreting of the Statistical Division of Hoover’s Food Ad- the effects of war on North Sea fisheries. ministration Program had impressed on him the In the early 1920s D’Ancona explained his constraints on human populations. In 1920 he observations to his future father-in-law, Vito published a successful prediction of the next Volterra. Volterra was then Professor of Math- US census using a mathematical model of growth ematical Physics in Rome, and D’Ancona’s data that he had rediscovered, the now famous logistic rekindled his interest, dating back to 1901, in the equation. possibilities of applying mathematics to biology. Pearl’s modelling was picked up by Johan Hjort Others had pursued those possibilities while during the First World War, while he studied in History of Fisheries 75 self-imposed exile in England. He had resigned MSY (see Schnute and Richards, Chapter 6, this as Director of Fisheries in Norway in protest volume, for modern developments). that secret fisheries negotiations he had assisted Graham immediately picked up on Hjort’s ap- with were not made public, as had been promised. proach, applying it to the data from the Great Fish- Following his repatriation to accept a professor- ing Experiment (Graham 1935). By fitting Pearl’s ship at the University of Oslo after the war, Hjort logistic equation to the data from and subsequent began to focus on the problems of the developing to the First World War, Graham showed that post- whaling industry (Hjort 1933). Noting the variabil- war fishing had reduced the population below the ity in the recruitment of young fish that he level giving largest catches, just as his Law of had demonstrated in 1914, Hjort was attracted to Fishing predicted would happen. But based on W. F. whales because ‘the fate of the stock is bound up Thompson’s success in managing the Pacific with the fate of a limited progeny’. Thus, he ex- halibut fishery, he went on to argue that catches pected recruitment to the population to be less could be increased if fishing effort was reduced. variable than for fish, and more amenable to study While Graham’s prediction from Hjort’s model (Hjort et al. 1933): contributed to the 1937 London Fisheries Agree- ment, the prediction was destined to be tested For studies of population these conditions again by war, rather than by fishery management. have the advantage of affording larger oppor- tunities of watching the new generation from birth onward, and thus obtaining defi- nite figures of the numerical strength of the 4.6 POST-SECOND different stages or year classes. WORLD WAR

By 1933 Hjort and his students demonstrated how The approaches taken by Baranov, W.F. Thompson, to use models like Pearl’s logistic equation to de- Hjort, Graham, and Hulme were extended imme- scribe the effects of harvesting on a whale popula- diately following the Second World War. Hulme’s tion. They focused on the nearly direct dependence young coauthors, Beverton and Holt, developed of recruitment on population size and the limi- his approach by including a more realistic model tation to recruitment by food resources at high for individual fish growth, proposed by Ludwig population sizes. Pearl had seen this as an inherent von Bertalanffy, and also structured the analysis in limitation, and postulated the existence of a maxi- terms of changing age of the youngest fish caught mum population size, the carrying capacity of the as well as the fishing intensity (Fig. 4.5). This habitat. One essential point in Hjort’s analysis was kind of analysis provides one of the mainstays of that any sustained harvest of a population resulted the modern yield-per-recruit forecasting (Shepherd in the population decreasing in size, as Garstang and Pope, Chapter 8, this volume). In this form the had demonstrated in 1900. A second implication of possibility of more than doubling the catch over the logistic model of fishing was that as the popula- time simply by increasing mesh sizes to protect tion size decreased due to fishing, the sustainable younger fish can be seen graphically. Although the catch would at first increase. However, as the fish- regulation of fishing intensity in the North Sea ing intensity continued to increase and the popula- was not likely, control of mesh size was allowed tion continued to decrease, there would be a point under the terms of the Overfishing Convention beyond which the sustainable catch would begin that arose out of the 1946 International Overfish- to decline, again just as Graham’s Law of Fishing ing Conference. This analysis, especially as suggested. One important concept that arose from codified and extended in Beverton and Holt’s 1957 Hjort’s model was that there was in fact a maxi- monograph ‘On the dynamics of exploited fish mum catch that could be sustained, a level even- populations’, became the basis for most manage- tually termed the maximum sustainable yield or ment attempts in the North Sea after the war. 76 Chapter 4

15.0

14.0

13.0

Fig. 4.5 Beverton and Holt’s 12.0 graph showing the change in pre- B' dicted North Sea plaice catches 11.0 (1000s of tonnes, solid lines show 115 equal catches as labelled along 10.0 110 lower right end of lines) for combi- A' 105 nations of fishing intensity (in-

(years) 100 stantaneous fishing mortality

r¢ 9.0 t 95 rate, F) and minimum age of the

90 fish harvested (tr¢ years). Predicted 8.0 85 catches were highest for mini- 80 mum ages of harvest around 11 7.0 75 years and for fishing mortality greater that 1.0, decreasing for 70 lower or higher minimum ages 6.0 65 and for lower fishing mortality. 60 Before the Second World War the 5.0 55 minimum age was 3.7 years and 50 the fishing mortality around 0.7 4.0 (solid circle labelled P). The catch 3.7 was 55000 tonnes, roughly half of BA P the 110000 tonnes predicted for 3.0 that fishing mortality rate if fish 2.5 0 0.5 1.0 1.5 8 were to be allowed to live at least nine years. (Source: Beverton F 1952.)

Similarly Hjort’s use of the logistic model was technique have led to modern-day surplus produc- picked up by Milner Schaefer in his 1954 paper tion models (Schnute and Richards, Chapter 6, this ‘Some aspects of the dynamics of populations volume). In 1960 four scientists were invited important to the management of the commercial by the then failing International Whaling Com- marine fisheries’. He developed a method for fit- mission to apply modern fishery population ting the model using fisheries statistics, and ap- dynamics methods to the Antarctic blue whale plied it to data from several fisheries, notably to fishery. These scientists, Douglas Chapman, the then recently defunct California sardine fish- Kenneth Allen, Sydney Holt and John Gulland, ery. He showed that by the end of that fishery in the used the Schaefer Curve to show that the Antarctic late 1940s the population had been reduced to a blue whale had been decimated from an initial fraction of its original size, and that the sustainable 210000 whales because its sustainable catch had catch was far less than the catches in the years be- been consistently exceeded for decades (Fig. 4.6) fore the fishery collapsed. The ‘Schaefer Curve’ (Chapman 1964; Small 1971). would eventually be applied to Antarctic whaling, The population models elaborated just after the which Hjort had focused on when he first devel- Second World War did not allow for some impor- oped the approach. Further refinements to this tant behaviour of fish populations and fisheries. History of Fisheries 77

15 1000 + 900 800 700 =0.1 + 10 1946/47 M 600 r– + 500

+ + 400

Catches Cases of salmon 300 5 + War years + Pre-war 200 + mean + + 100 + + + + 0 +++ 1885 1890 1895 1900 1905 1910 1915 ++ +++

Surplus stock = sustainable yield (thousands) 0 0 50 100 150 200 Year Stock (thousands) Fig. 4.7 Graph of the numbers of cases of Fraser River sockeye salmon canned from 1885 to 1917, reflecting Fig. 4.6 Chapman’s graph of the predicted sustainable the extremely high catches every fourth year (e.g. 1901, yield of Antarctic blue whales (vertical axis) at different 1905, etc.) until 1913. The catch four years later (in 1917, stock sizes (horizontal axis), showing predicted yields at rightmost data point) was substantially below the different stock sizes (¥) and actual catches in successive average fourth-year catches, and was blamed on rock- years (•) beginning in the 1946/47 Antarctic season. The falls from railroad construction blocking the Fraser theoretical maximum sustainable yield (peak of dashed River at Hell’s Gate in 1913. (Source: from Smith 1994, line) was roughly 6000 whales, and could be obtained reprinted with permission from Cambridge University only when the population size was 110000 animals. By Press.) the Second World War, the (unsustainable) yields were larger than that maximum and the stock size had been catches would be possible with less fishing effort reduced to less than half of the required level, showing that blue whales had already been heavily overfished. (see Hannesson, Chapter 12, this volume). In the (Source: from Chapman 1964.) same manner, the importance of Lankester’s 1884 concerns about ‘complex interactions’, although demonstrated by Garstang and repeatedly by oth- For example, no matter how one tried, simple ers, continued to be ignored. Finally, Kitawara’s equations such as the logistic just could not mimic and Uta’s fisheries oceanography continued to the cyclic behaviour of D’Anconna’s data or play a large role in the ever-expanding distant the four-year oscillations of the sockeye salmon water fleets, but made virtually no contribution to (Oncorhynchus nerka) fishery in the Fraser science directed at the control of fisheries. River in British Columbia. Volterra’s and Lotka’s approaches, while influential in terrestrial ecology 4.6.1 Oscillations (Kingsland 1985), were not picked up despite Volterra’s assurance that his approach would Predictable four-year oscillations in the Fraser provide ‘an easy way of calculating the maxi- River sockeye catches (Fig. 4.7) were early sug- mum output of fisheries’ (Volterra 1926). Modern gested to be due to differences in productivity of developments dealing with interactions between four separate populations of fish using the river, species can be found in Shepherd and Pope (Chap- each returning to spawn after four years at sea. ter 7, this volume) and Pauly and Christensen In 1914, when the upstream migrating salmon (Chapter 10, this volume). Similarly, the models reached the notorious Hells Gate passage, they adopted by fishery biologists did not account for encountered a huge blockade of rocks from the the economics of fishing, and especially the clearly construction of the second Canadian transconti- demonstrated behaviour of fisheries to reduce fish nental railway. Despite heroic efforts to clear the populations to low levels even though higher blockade, there were fears that this, the largest of 78 Chapter 4 the four populations, would be wiped out. These accumulated since 1912, William Herrington had fears were confirmed when the pattern of the noticed that the winter abundance of three-year- catches abruptly changed four years later. The old haddock tended to be low when the winter 1917 catch was much lower than expected, only a catches of adult haddock three years earlier were fraction of 1914, confirming the biologists’ expla- either high or low. When the adult catches were nation for the oscillations. intermediate, the corresponding three-year-old After the war William Ricker took up this ques- abundance was markedly higher. Herrington had tion, drawing on a study of sockeye salmon in also noticed that when the abundance of adult had- Cultus Lake, British Columbia, that he had begun dock was high, adult fish tended to expand their his career with. That study had been prompted as range into that usually occupied only by young part of a Canadian review of the effectiveness of its fish. He suggested that this spatial expansion may salmon hatchery programme. The focus was on increase competition with the young, to their the growth and survival of young salmon over the detriment. year they are in the lake, before they migrate down Ricker in 1954 reanalysed Herrington’s data, the Fraser River. Analyses of those data convinced plotting the abundance of the young against Ricker that part of the cause of the four-year the corresponding abundance of their parents pattern, which was intriguingly similar to cycles (Fig. 4.8). This approach was widely adopted, and in catches of fur animals that Elton (1927) had biologists began to comb through fishery statistics shown from the Hudson Bay Company, was looking for evidence of changes in population lower individual fish growth rates. He termed this vital rates with population density that might phenomenon ‘cyclic dominance’ (Ricker 1954), and explain the variability in recruitment that Hjort argued that it occurred because slower growing had identified in 1914 and the regular oscillations fish were smaller and hence more vulnerable to seen in some fisheries (see Myers, Chapter 6, predation for a longer time. Similarly, there was Volume 1). the possibility that there would be more predators in the lake if the previous year’s salmon hatch had 4.6.2 Bioeconomics been larger. Ricker published his analysis in 1950 under Modern theories of fisheries bioeconomics, which the title ‘Cyclic dominance among Fraser [River] are reviewed by Hannesson (Chapter 12, this vol- sockeye’. The same year Lamont Cole, a terrestrial ume) can be traced back to 1905, when H. M. Kyle biologist, published a controversial paper titled had argued that the economics of fishing was ‘Population cycles and random oscillations’. He the essence of the problem of overfishing. The con- raised questions about the reality of some apparent trolling factors were capital invested and income cycles in abundance, claiming that some of the earned, and, along with population productivity, cyclic data could not be distinguished statistically growth rate and mortality, defined the problem. from random numbers. The response to Ricker’s Kyle did not consider capital and earnings to be and Cole’s papers was the identification of several proper scientific issues, however, because they are mathematical models that, unlike the simple not ‘subject to natural law’. Further, he argued, logistic, could generate the observed oscillations. even if they were scientific issues, as people like Some of these involved within-year dynamics Alfred Marshall in his book Principles of Econom- such as delayed growth or cannibalism, as Ricker ics were beginning to argue (Marshall 1890), the had suggested, while others involved time lags in calculation of ‘the precise point where overfish- biological processes. ing begins’ would be a waste of time because the Ricker expanded his analysis and considered answer would change if the price of fish changed other fishery data as well in his 1954 paper ‘Stock by even a ‘fraction of a penny’. and recruitment’. One interesting case was the Kyle’s argument prevailed, and biologists and Georges Bank haddock. Analysing fishery data fishery managers steered clear of the economics History of Fisheries 79

cused for the last half century. More generally, 22 however, Gordon argued that only by including 14 fishermen as part of the ecosystem, including both their economics and their behaviour, could the 12 problem of overfishing be addressed. He noted that 15 (Gordon 1953)

13 practically all control measures have . . . been designed by biologists, with the sole at- 12 tention paid to the production side of the 8 24 problem and not to the cost side. The result

Recruits 23 has been a wide-open door for the frustration 40 16 21 of the purposes of such measures. 33 39 25 The adoption of Gordon’s perspective has come 4 38 36 29 only slowly and partially in fisheries, and more 34 41 20 35 32 recent developments are mentioned in Hart and 30 31 Reynolds (Chapter 1, this volume). 37 26 42 28 27 17 18 19 0 4.6.3 Complex interactions: 0 204060 Adults predators, prey and the ecosystem The importance of Lankester’s ‘complex in- Fig. 4.8 Ricker’s graph of the relationship between winter abundance of spawning haddock (adults, thou- teractions’ was first demonstrated by Garstang sands of lbs caught per day fished) and winter abundance analysing the Scottish experiment that resolved of the fish resulting from that spawning three years later the Huxley–Lankester debate. Similarly, while (thousands of lbs caught per day fished). The the ICES interpretation of the Great Fishing Ex- diagonal line indicates the case when the abundance of periment was in terms of a single species – plaice – three-year-old fish would be equal to that of their parents the interpretation by D’Ancona and Volterra in in the year the young fish were spawned. The domed terms of predators and prey was ignored. The sim- curve depicts Ricker’s theoretical density-dependence model of how spawning success decreases with increas- ple logistic model, which considered the relation- ing size of the spawning stock. Ricker showed that his ships with other species only in an aggregate and model fits well to the earlier data (solid circles) when static manner by assuming unvarying carrying spawning stock was more abundant (solid line), but not capacity and maximum sustained yield, was in- to the later data (open circles) when it was less abundant adequate for these issues. The effect of predator (dashed line). (Source: Ricker 1954.) species on catches, identified as a problem at least as early as the 1880s in France, was also not ad- dressed. The potential for overfishing one of two of fishing in most countries. William Ricker ac- species harvested by the same fishery, which was knowledged the wisdom of that decision in a 1946 first identified by Lotka in 1925, was not addressed review of the North Pacific halibut fishery, even as even though it was obviously occurring in he acknowledged the increasing numbers of boats Antarctic whaling (Tonnessen and Johnsen 1982). attracted to an ever-shortening open season. The extreme depletion of blue whales demons- H. Scott Gordon took up the question again in trated by Chapman and colleagues was dependent a 1953 paper, agreeing with Kyle that overfishing on that fishery also harvesting the more abundant was primarily an economic problem and suggest- (albeit smaller) fin and sei (Balaenoptera borealis) ing that the problem had been too narrowly fo- whales. Without the profits from those still abun- 80 Chapter 4 dant species, the whaling fleet could not have af- class strength in 1914, attempts to describe the ef- forded to continue to hunt the blue whale remnant fects of the environment on fishery recruitment to near extinction. have largely failed (see Myers, Chapter 6, Volume Moving from a multispecies perspective to the 1). Symposia were held in 1951 and again in 1959 even more complex ecosystem perspective, the to address ‘[t]he question of the dependence on detailed process of modelling of ecosystem dyna- milieu-factors of the life in the sea’. Over the next mics undertaken under the International Biological decade controversy raged over the content and Program was not quickly taken up by fisheries sci- value of such a discipline (Chapman 1962; Kendall entists. Holt, by then a leading fisheries scientist and Duker 1998). Some argued that such studies working for the United Nations Food and Agricul- were essential if the cause of variability is to be tural Organization (FAO), reportedly discouraged understood, while others argued that they were a fisheries involvement in that approach during a waste of time. formative meeting with the International Union Despite the controversy, fisheries oceanogra- of Biological Sciences in 1962. Holt argued that phy developed as a discipline, eventually acquiring ‘fishery biology was very largely internationalized its own journal by that title. However, the causes and did not really need much more international of year-class variability are still not completely organization – and at any rate, if it did, this should known (Myers, Chapter 6, Volume 1), and the ef- fall to the province of FAO’ and not to the Interna- fects of this variability have been included in tional Union of Biological Sciences (Waddington fishery population models to only a limited degree. 1975). The utility of the fishery population models elabo- There was, however, developing ecological in- rated in the early 1950s without an understanding terest in some fisheries after the war. In 1949, for of environmental effects was challenged by Uda in example, a multispecies trawl survey of Georges 1962, when he noted that ‘[a]fter the World War II, Bank was initiated. Although continued only until the estimation of fisheries yield or the absolute 1952, a similar survey was resumed in 1963 and population size on vital statistics, using the power- continues today as the backbone of the US Govern- ful tool of population dynamics have made some ment fisheries laboratory in Woods Hole (Smith progress. However, [some scientists were] . . . 2002), where Baird first initiated his broad-scope convinced soon that the reproduction potentials research programme in 1871. The development of or recruitments are too much changing in response ecosystem perspectives and methods in fisheries is to the change of natural environments’. described by Pauly and Christensen (Chapter 10, this volume) and multispecies methods in stock assessment are described by Shepherd and Pope 4.7 CONCLUSIONS (Chapters 7 and 8, this volume). The effect of environmental changes on fishes, The application of science to fisheries problems, although discussed since before the beginning of beginning in the 1850s, has been fruitful. The un- ICES, was not a main thrust of fisheries research derlying mechanisms for fluctuations in fishery prior to the Second World War. Work in Japan had catches have been elucidated, at least to one level. demonstrated the usefulness of fisheries oceano- The several dimensions inherent in the term graphy for answering the same operational ques- overfishing have been understood, ranging from tions that St Paul had dealt with on Lake Kinneret recruitment and growth overfishing, to economic so long ago. Further, Martin Burkenroad (1948) had overfishing. The basic principles from demogra- stridently raised the question of the effects of phy have been incorporated into the increasingly environmental variability in the late 1940s, espe- sophisticated population models, and manage- cially criticizing the basis for management of had- ment measures that would alleviate some aspects dock and halibut during a 1948 symposium. Despite of overfishing have been developed. But the lesson work since Hjort’s demonstration of variable year- from the history of fisheries science is that many History of Fisheries 81 problems that were identified in the 1800s and Baranov, F. (1926) On the question of the dynamics of the early 1900s remain unsolved. Some of these prob- fishing industry. Byulleten Rybnogo Khozyaistva lems have increasingly been addressed as the scope (1925) 8, 7–11. Beverton, R.J.H. (1952) Some observations on the princi- of fisheries biology has expanded, and as other ples and methods of fishery regulation. Unpublished previously excluded disciplines, such as oceano- manuscript. graphy and economics, have been integrated into Beverton, R.J.H. and Holt, S.J. (1957) On the dynamics of the study of fisheries. However, many of the issues exploited fish populations. United Kingdom Ministry that have not been systematically addressed by of Agriculture and Fisheries, Fisheries Investigations fisheries scientists are becoming increasingly im- (Series 2), 19. London: Her Majesty’s Stationery Office. portant. The bycatch of other species, such as Bigelow, H. (1931) Oceanography: its Scope, Problems and Economic Importance. Boston: Houghton Mifflin. sea birds and marine mammals, has only begun to Borley, J.O., Russell, E.S., Graham, M., Wallace, W. and be addressed (see Kaiser and Jennings, Chapter Thursby-Pellan, D.E. (1923) The plaice fishery and 16, this volume). Similarly, while many biologists the war: preliminary report on investigations. United speak of carrying capacity and maximum sus- Kingdom Ministry of Fisheries and Agriculture, Fish- tained yield of their species, few are attempt- eries Investigations, (Series 2), 5. London: MAFF. ing to address the joint effects of even all of the Burkenroad, M.D. (1948) Fluctuation in abundance of Pacific halibut. Bulletin Bingham Oceanographic fish species, much less non-fish species, on the Collection 11, 81–129. ecosystem (but see Pauly and Christensen, Chap- Chapman, D.G. (1964) Final report of the Committee of ter 10, this volume). Similarly, fisheries have Three Scientists. Report International Whaling Com- frequently been observed to serially deplete tar- mission 14, 39–92. get species over time. The example of the Chapman, W.M. (1962) Comments on fishery oceanogra- Atlantic halibut in the Gulf of Maine mentioned phy. In: W.M. Chapman (ed.) Comments on Fishery above can be multiplied with many other species Oceanography, Volume 1. Prepared for the Working Party on Fishery Oceanography of the Scientific (Pauly et al. 1998). The cumulative effect of long- Committee on Oceanic Research of the International term fisheries can reshape ecosystems without Council of Scientific Unions, 20 August 1962, pp. 1–32. that being recognized because the baseline that Cleghorn J. (1854) On the fluctuations in the herring fish- biologists too often compare to is the ecosystem eries. British Association for Advancement of Science as they first encountered it (Pauly 1995; Holm 24, 124. 1996; Jackson 1997; Holm et al. 2002). These areas Cole, L.C. (1950) Population cycles and random oscilla- will have to be addressed by the 21st century fish- tions. Journal of Wildlife Management 15, 233–52. Dannevig, G.M. (1910) The utility of sea-fish hatching. eries scientists. Fisheries Bulletin 28, 811–16. Dunlop, J. (1978) The British Fisheries Society 1786– 1893. Edinburgh: Donald. REFERENCES Elton, C.S. (1927) Animal Ecology. New York: Macmillan. Anon. (1921) The history of trawling: its rise and devel- Engholm, B. (1961) Fishery conservation in the Atlantic opment from the earliest times to the present day. Ocean. In: G. Borgstrom and A.J. Heighway (eds), The Fish Trades Gazette & Poultry, Game & Rabbit Atlantic Ocean Fisheries. London: Fishing News Trades Chronicle 19 March, odd-numbered pages Books, pp. 40–8. 21–69. Garstang, W. (1900) The impoverishment of the sea – a Baird, S.F. (1872) Report on the conditions of the sea fish- critical summary of the experimental and statistical eries, 1871. Report of the United States Fish Commis- evidence bearing upon the alleged depletion of the sion 1, vii–xli. trawling grounds. Journal of the Marine Biological Baird, S.F. (1887) Some opinions of the importance of arti- Association 6, 1–69. ficial propagation. Report of the United States Fish Goode, G.B. and Collins, J.W. (1887) The fresh halibut Commission 8, lxxiv–lxxviii. fishery. In: G.B. Goode (ed.) The Fisheries and Fishing Baranov, F.I. (1918) On the question of the biological basis Industries of the U.S., Section V: History and Methods of fisheries. Nauchnyi Issledovatelskii Ikhtiologich- of the Fisheries. Washington, D.C.: Government eskii Institut, Izvestiya 1, 81–128. Printing Office, pp. 3–89. 82 Chapter 4

Gordon, H.S. (1953) An economic approach to the opti- Kyle, H.M. (1905) Statistics of the North Sea fisheries. mum utilization of fishery resources. Journal of the Part II. Summary of the available fisheries statistics Fisheries Research Board of Canada 10, 442–57. and their value for the solution of the problems of over- Graham, M. (1935) Modern theory of exploiting a fishery, fishing. Rapports Conseil Internationale pour la Ex- and application to North Sea trawling. Journal du Con- ploration de la Mer 3. seil 10, 264–74. Lankester, E.R. (1884) The scientific results of the exhibi- Graham, M. (1943) The Fish Gate. London: Faber. tion. Fisheries Exhibition Literature 4, 405–45. Heincke, F. (1913) Investigations on the plaice. General Lotka, A.J. (1925) Elements of Physical Biology. Report 1. The Plaice Fishery and Protective Regula- Baltimore, Md.: Williams and Wilkins. tion, Part 1. Rapports Conseil Internationale pour la Marshall, A. (1890) Principles of Economics. London: Exploration de la Mer 17, 1–153. Macmillan. Herrington, W.C. (1943) Some methods of fishing man- Nesbitt, R.A. (1943) Biological and economic problems of agement and their usefulness in a management pro- fishing management. United States Fish and Wildlife gram. United States Fish and Wildlife Service, Special Service, Special Scientific Report 18, 23–53, 61–8. Scientific Report 18, 3–22, 54, 58. McIntosh, W.C. (1899) The Resources of the Sea [2nd Hjort, J. (1914) Fluctuations in the great fisheries of edn., 1921.] Cambridge: Cambridge University Press. northern Europe. Rapports Conseil Internationale Pauly, D. (1995) Anecdotes and the shifting baseline pour la Exploration de la Mer 20. syndrome of fisheries. Trends in Ecology and Evolu- Hjort, J. (1933) Whales and whaling. Hvalradets Skrifter tion 10, 430. 7, 7–29. Pauly, D., Christensen, V., Dalsgaard, J., Froese, R. and Hjort, J., Jahn, G. and Ottestad, P. (1933) The optimum Torres Jr., F.C. (1998) Fishing down marine food webs. catch. Hvalradets Skrifter 7, 92–127. Science 279, 860–3. Hjort, J. and Lea, E. (1914) The age of herring. Nature 94, Petersen, C.G.J. (1903) What is overfishing? Journal of 255–6. the Marine Biological Association 6, 587–94. Holm, P. (1996) Catches and manpower in the Danish Peterson, E. (1993) The Message. Colorado Springs: fisheries, c.1200–1995. In: Poul Holm, David J. Starkey NavPress. and Jón Th. Thór (eds) Studia Atlantica 1: North Ricker, W.E. (1950) Cycle dominance among Fraser sock- Atlantic Fisheries, 1100–1976. National Perspectives eye. Ecology 31, 6–25. on a Common Resource. Esbjerg: Fiskerimuseets Ricker, W.E. (1954) Stock and recruitment. Journal of the Forlag, pp. 177–226. Fisheries Research Board for Canada 11, 559–623. Holm, P., Starkey, D.J. and Smith, T.D. (2002) Introduc- Rozwadowski, H. (2002) The Sea Knows No Boundaries: tion. In: Poul Holm, Tim Smith and David J. Starkey A Century of Marine Science under International (eds) Exploited Seas: Directions for Marine Environ- Council for the Exploration of the Sea. Seattle: Univer- mental History, Research in Maritime History 20, sity of Washington Press. in press. St Johns, Newfoundland: International Sahrhage, D. and Lundbeck, J. (1992) A History of Fish- Maritime Economic History Association, Memorial ing. Berlin: Springer-Verlag. University. Schaefer, M.B. (1954) Some aspects of the dynamics Hulme, H.R., Beverton, R.J.H. and Holt, S.J. (1947) of populations important to the management of the Population studies in fisheries biology. Nature 159, commercial marine fisheries. Inter-American Tropical 714–15. Tuna Commission Bulletin 1, 27–56. Huxley, T.H. (1884) Inaugural Address. Fisheries Exhibi- Schevill, W.E. (ed.) (1974) The Whale Problem. tion Literature 4, 1–22. Cambridge: Harvard University Press. Innis, G. (1940) The Cod Fishery. Toronto: University of Sinclair, M. and Solemdal, P. (1988) The development of Toronto Press. ‘population thinking’ in fisheries biology between Jackson, J. (1997) Reefs since Columbus. Coral Reefs 16 1878 and 1930. Aquatic Living Resources 1, 189–213. (Supplement 1), S23–S32. Small, G.L. (1971) The Blue Whale. New York: Columbia Johnstone, J. (1905) British Fisheries. London: Williams University Press. and Northgate. Smith, T.D. (1994) Scaling Fisheries: the Science of Mea- Kendall, A.W. and Duker, G.J. (1998) The development suring the Effects of Fishing, 1855–1955. Cambridge: of recruitment fisheries oceanography in the United Cambridge University Press. States. Fisheries Oceanography 7, 69–88. Smith, T.D. (2002) The Woods Hole bottom trawl re- Kingsland, S.E. (1985) Modeling Nature. Episodes in the source survey: development of fisheries-independent History of Population Ecology. Chicago: University of multispecies monitoring. In: E. Anderson (ed.) 100 Chicago Press. Years of Science under ICES. A Symposium held in History of Fisheries 83

Helsinki, 1–4 August 2000. ICES Marine Science Sym- Uda, M. (1962) The accomplishments of fisheries posia 215: in press. Copehagen: International Council oceanography in Japan and its future oceanography. for the Exploration of the Seas. In: W.M. Chapman (ed.) Comments on Fishery Smith, T.D., Gjøsæter, J., Stenseth, N.C., Kittilsen, M.O., Oceanography, Volume 3. Prepared for the Work- Danielssen, D.S., Solemdal, P. and Tveite, S. (2002) A ing Party on Fishery Oceanography of the Scientific century of manipulating recruitment in coastal cod Committee on Oceanic Research of the International populations: the Flødevigen experience. In: E. Anderson Council of Scientific Unions. 25 December, 1962, pp. (ed.) 100 Years of Science under ICES. ICES Marine 46–56. Science Symposia 215, in press. Copenhagen: ICES. Uda, M. (1972) Historical development of fisheries Taylor, J., III. (1999) Making Salmon: An Environmental oceanography in Japan. Proceedings Royal Society History of the Northwest Fisheries Crisis. Seattle: Uni- Edinburgh (B) 73, 391–8. versity of Washington Press. Volterra, V. (1926) Fluctuations in the abundance of Thompson, W.F. and Bell, F.H. (1934) Biological statistics a species considered mathematically. Nature 118, of the Pacific halibut fishery. 2. Effect of changes in in- 558–60. tensity upon total yield and yield per unit of gear. Re- Waddington, C.H. (1975) The origin. In: E.B. port of the International Fisheries Commission 8. Worthington (ed.) The Evolution of IBP. Cambridge: Tonnessen, J.N. and Johnsen, A.O. (1982) The History of Cambridge University Press, pp. 4–12. Modern Whaling, trans. R.I. Christophersen. Stanford, Wimpenny, R.S. (1953) The Plaice. London: Edward Ca.: University of California Press. Arnold. 5 Gathering Data for Resource Monitoring and Fisheries Management

DAVID EVANS AND RICHARD GRAINGER

ent international instruments applied to inland 5.1 FISHERIES fisheries, mainly because they are enclosed INFORMATION FRAMEWORK within countries or shared between two or more 5.1.1 Information for a fisheries neighbouring states, but in these cases there are knowledge system now numerous bilateral or multilateral agree- ments regulating these fisheries, such as for the Since 1950 global fisheries have come under ever- North American Great Lakes and Lake Victoria. increasing exploitation pressures caused by rising In addition, the international community has demand, technological innovation and efficiency, adopted a voluntary instrument, the Code of Con- and the increasing globalization of markets. World duct for Responsible Fisheries (Code) (FAO 1995), production has increased from about 20 million to bridge the gap between international political tonnes then to more than 100 million tonnes today acceptance of the needs for sustainability and the (FAO 1996a; see Hart and Reyholds, Chapter 1, actual conduct of fisheries. In both UNCLOS and Volume 1). It has been argued that this level is ap- the Code their articles exhort states, organizations proaching the maximum that can be obtained from and individuals to collect data, develop knowledge nature. The reach of fishing fleets into almost and apply appropriate measures. all the distant and deep oceans, the utilization of In this chapter we look at the nature and more species, more intensively, and calamitous extent of the information required to populate a collapses of important fisheries have provided im- ‘knowledge system’ from which the three inter- portant evidence that this may be true. Sustain- linked fishery management domains of Policy, ability and a clear understanding of the pitfalls of Planning and Implementation can draw (see Fig. overexploitation have become the driving forces 5.1) of international fisheries research and fishery management. Knowledge about all the compo- 5.1.2 Fishery information domains nents of fishery systems has become an imperative alongside institutional mechanisms to put that There are many ways in which the information do- information into practice. The primary instru- mains within the fisheries sector might be catego- ment for improved fisheries management is the rized. For our purposes we have chosen to group United Nations Convention on the Law of the Sea very wide ranges of information within four main (UNCLOS 1982), through which the global fish- domains. These are addressed in more detail in eries community has begun to bring a more ratio- subsequent sections and many are dealt with ex- nal and sustainable approach to the exploitation of tensively throughout these volumes in relation to wild marine fisheries. There have been no equival- specific issues: Gathering Data for Monitoring and Management 85

K N Policy O Fishery and operations W L E D Biology and environment Planning G E Economic and financial S Y S Implementation T Sociocultural E M Fig. 5.1 Schematic representation Management domain Information domain of a fisheries knowledge system.

• Fishery operations: this includes all aspects of social, economic and geographic diversity. There the primary capture and processing of aquatic re- are common features of fish stock dynamics sources. Who caught what amounts of fish, when and assessments that generally apply universally, and how? What operational management informa- even across the thousands of species and stocks tion is required? targeted by fisheries. But the conduct of fisheries, • Biology and environment: this includes all char- even on single stocks, generates an array of inter- acteristics of the biology, dynamics and interac- acting influences that is both hard to define and tions of exploited living aquatic resources and difficult to measure, and may well be unique to their associated species within the biological and any one fishery. For example, purse seiners from physical framework of the ecosystem. one country may target a transboundary coastal • Economic and financial: this includes all aspects pilchard stock in areas beyond the reach of small- of the conduct of the fisheries sector, including scale fishers in another country who fish by beach human resource, financial, economic, production seine. and trade issues. Increasingly fishery managers acknowledge the • Sociocultural: this includes all features that are need to include politico-socio-cultural, economic needed to evaluate fisheries policy and manage- and environmental issues within the frameworks ment (including institutional mechanisms) in re- that they must consider when addressing sus- lation to the distribution of resources, income and tainable utilization and management decision food, equity and demographics, and participant de- making. It is no longer the case that fishery pendence and social status. management focuses mainly on the issues of how much fish may be taken or how much fishing effort 5.1.3 Information and may be applied. Of course, these primary issues fisheries management define, however imprecisely, the removals from stocks that will maintain sustainability, or All these information domains are required for progress towards it, and thus support the biological fisheries management in its widest sense within basis of the fishery economic system on which par- the political and economic systems of states and ticipants depend. Issues of fishing rights, access their regions. The range of issues and the data re- and tenure also come within the management quired for analysis and decision making within equation, as does the whole subject of minimiza- these four domains is enormous. Partly this is tion of environmental deterioration from the ef- due to ecosystem diversity but mostly it is due to fects of fishing or pollution. 86 Chapter 5

5.1.4 Information for fisheries policy, 1995; Gabriel and Mace 1998). Many of the follow- the precautionary approach and ing chapters in this book address the analyses reference points and models required to estimate these reference points, in particular the chapter by Sparre and Hart There are three main areas of fisheries policy that (Chapter 13, this volume). circumscribe the ways in which fishery managers conduct their activities: contributions of fisheries 5.1.5 Fisheries information cycle to food security, contributions of fisheries to the economy, and the effects of fishing on the environ- Given that national and subnational fisheries poli- ment. In all cases, the Code recommends a precau- cies are derived from a consideration of all the bio- tionary approach: precaution to ensure that fish logical, social and economic circumstances that removals for food and other uses are consistent are particular to a country, it follows that the sets with sustainability; precaution that the use of re- of data required to inform a national policy will newable aquatic resources as an economic activity also be unique. To enable policy to be turned into is maintained to prevent economic and social dis- fisheries management advice that works (i.e. satis- location and to approach optimum utilization; and fies the goals of policy, responds to environmental precaution to prevent environmental degradation, and stock variations and changes in economic including biodiversity. development), there always needs to be an in- Precaution requires the application of prudent formation cycle that continuously supplies the foresight, which takes into account both best knowledge system (see Fig. 5.2). At the outset, available scientific evidence and the uncertainties policy and its objectives need to be interpreted in of fisheries systems when knowledge is incom- terms of performance indicators such as biological plete. Fisheries policy is not specifically dealt with reference points, against which the results of fish- in this book, but it is clear that precaution as an eries management are measured. For example, accepted policy precondition is becoming more fishery managers might ask, what is the current firmly established. The Technical Consultation level of fishing mortality, a performance indicator, on the Precautionary Approach to Capture Fish- against the fishing mortality that satisfies sustain- eries and Species Introductions (FAO 1996b) elabo- ability which is a current precautionary reference rated four key areas in which the precautionary point, either at the limit to biological productivity approach needs to be applied: fisheries manage- or at a target during stock rebuilding? ment, fisheries research, fisheries technology and Sometimes fishery performance indicators may species introductions. For each one of these, each be constructed from single variables, such as total containing numerous underlying issues, the na- catch, but in most cases they will be a combination ture of appropriate precaution needs to be estab- of variables for which arrays of information need lished in the form of operational targets and to be collected, for example fishing mortality. In constraints. Much recent work has been con- some cases multivariable indicators may be sim- ducted on what this means in terms of fisheries man- ple combinations of variables; in others, such as agement, less so for the other areas. For example, for fish populations, the variables may be related the basis for the measurement, and hence the data in complex mathematical models (see Chapters requirements, of fishing capacity and fishing tech- 6–13, this volume). Whatever the case, planning nology efficiency is still in its early development. for the collection of the data necessary for these For fish stock management there has been variables will involve determining their feasibil- significant progress to specify how operational ity, priority, frequency, quality and quantity, and targets and constraints should be measured, whether standards for these can be adopted or de- particularly through the definition of biological rived. Once a performance indicator and its vari- reference points, including target and limit or ables are defined according to these determinants, threshold reference points (Caddy and Mahon the implementation phase begins through the de- Gathering Data for Monitoring and Management 87

General policy decisions • Identify policy and management objectives • Incorporate the precautionary approach

WHY Objectives of data collection • Link information needs to policy and management • Identify types of indicators which will be required • Determine any regional requirements

Performance indicators and observable data variables • For each variable, determine whether it can be Analytical collected, its priority, its collection frequency, the methods required data quantity, quality and WHAT standardization needed. • Link the required indicators to variables for fishing/operational, biological, economic and sociocultural assessments. Feedback Data collection strategy • Obtain baseline information for design Logistics • Decide on complete enumeration or sampling and • Review operational constraints resources Data collection methods • Decide how the variable will be recorded • Link the required variables to their possible sources and collection methods

Data management • Design the database HOW • Plan database operations and maintenance • Plan data access and dissemination

Planning and implementation • Planning phase: legal, institutional and budget issues • Implementation phase: training, monitoring and administration issues

System appraisal Fig. 5.2 The fisheries information • Update and improve the programme using cycle. (Source: from FAO 1999.) experience and feedback

velopment of a strategy, the definition of data col- Finally, a system of monitoring is required to deter- lection methods and the means to maintain it, and mine whether all this effort actually both satisfies the establishment of the legal, institutional and proper calculation of the performance indicator budgetary environments in which it will operate. and ensures that the performance indicator is ap- 88 Chapter 5 propriate to the evaluation of the effectiveness Where possible the use of standardized and inter- of the management measures. In some cases this nationally recognized units of measurement, may be relatively simple but for multivariable particularly for dimensions and mass, and indicators simulation modelling can assist this nomenclature should be adopted. The global net- feedback process by revealing both the sources of work that forms the Coordinating Working Party uncertainty and inadequacy in the whole proce- on Fishery Statistics (FAO is the Secretariat) devel- dure, and the risk or probability that these sources ops much of this work. The following interna- of uncertainty and inadequacy are important or tional classification systems form the basis for significant. many useful standards: At each level in the fisheries information cycle • Species: International Standard Statistical Clas- there will be many decisions, actions and the appli- sification for Aquatic Animals and Plants or the cation of resources to continuously improve the FAO 3-alpha species codes developed for commer- knowledge base. A few examples are given in Fig. cial fish. 5.2. The advantages of this are clear. Even within • Fishing vessels: International Standard Statisti- the framework of the precautionary approach, calClassificationfor Fishery Vessels (FAO 1985). ‘States shall be more cautious when information • Fishing gear: International Standard Statistical is uncertain, unreliable or inadequate. The ab- Classification for Fishing Gear (FAO 1982). sence of scientific information shall not be used as • Products: Harmonized Commodity Description a reason for postponing or failing to take conserva- and Coding System (World Customs Organization, tion and management measures’ (United Nations 1994). 1995). Thus, investment in appropriate informa- • Oceanography: Committee for International tion will always reduce the degree of precaution re- Oceanographic Data Exchange. quired, and hence will enable a closer approach to Many other global organizations have developed the target or limit. Assuming that analytical meth- standards that may be important to the variables ods are accurate and interpretation correct then concerned, including the World Meteorological more information means more fish may be caught Organization (climate and weather measure- or fewer controls placed on fishing activity. ment), World Health Organization (nutritional and Put briefly, investment in data collection and health values), International Monetary Fund analysis will increase or sustain welfare and earn (economic and financial measures), and the Inter- revenues from fisheries, because it reduces the risk national Labour Organisation (human resource of overexploitation and leads to the establishment classifications). and management of sustainable exploitation pat- terns (FAO 1999). The precautionary approach can 5.1.7 Information precision be seen as a major incentive for the collection of re- and accuracy liable and appropriate fisheries and environmental information. Fisheries data and their analysis contribute signifi- cantly to the costs of national and regional fish- 5.1.6 Information standards eries management, often equivalent to more and classifications than 20% of gross value of the catch, in some cases higher (Larkin 1997). The precision of the data is Part of the planning process for information collec- thus important because, in general, the greater the tion, once the management objectives and indica- precision required, the more it will cost to ensure tors have been decided, requires definition of the its accuracy. For each data type and each situation data standards and coding to adopt for different it is necessary to define the precision required both classes of variables. The use of standards assists in for the satisfaction of the variable’s requirements a wide range of issues from data form design to and for instructions for data collection. Clearly database comparability and non-redundancy. measures of fish mass will be different for different Gathering Data for Monitoring and Management 89

fisheries; perhaps to the nearest tonne for indus- between the strata, which are then amenable to trial small pelagic fisheries, or to the nearest kilo- comparative analysis. The major strata in fisheries gram for recreational angling. Similarly, the are usually one or a combination of space, time, precision of dimensions needs to be established; to landings, vessels, gears, enterprises, trade, people, the nearest millimetre, say, for shrimp carapace the environment and the specific requirements of length or the nearest metre for water depth at-sea fishery-independent research. Subdividing measurement. the data population into minor strata, and then randomly sampling those, will enable further 5.1.8 Data sampling reduction of data variability. For example, a and stratification major stratum for fish trade might comprise markets/auctions, intermediaries/wholesalers/ Consideration of the coverage, location, quantity retailers, or exporters/importers as minor strata. and quality of data sampling also form part of the The key operational constraint of cost is also information cycle, generally as one of the final what drives stratification. The amount of effort, stages prior to the actual collection of data. At the defined as the number of samples, hence cost, falls beginning, or during revision of an information as bias is eliminated by stratification. At the out- programme, it is common to undertake fishery set, decisions on the quantity of data by stratum frame surveys or censuses in which complete enu- may often be made based on the results of cen- meration (100% coverage) of the basic structure of suses, but simulation can also be undertaken, par- the fishery sector is compiled, including produc- ticularly when revising sampling design based on tion, infrastructure, employment and community experience. In general, of course, the larger the dependence, and may also include environmental sample then the higher the accuracy. However, in- baselines. From these surveys, decisions can be creased accuracy is not proportional to sample made whether to maintain complete enumeration size, hence to the cost or effort applied. Depending or conduct sample surveys in such a way that the on the purposes of the data, there may well be a estimate from the data samples is as close as possi- cost/accuracy combination that satisfies both op- ble to the true value for the data population. The erational and statistical constraints. Even during difference between these two is the bias of the esti- sampling operations, such as very expensive trawl mate, and it is extremely important to calculate or bioacoustic surveys, it is usually wise to con- and account for this bias by statistical methods. tinuously estimate bias and to curtail sampling Reducing bias can be undertaken by increasing (hence costs) once a minimum satisfactory bias is sample size but also in other ways, partly by ran- reached. dom sampling to avoid sampling error and partly at Poor data quality, even assuming sample pre- the design stage of the programme. Sampling, of dictability from stratification, can limit any sam- course, takes much less effort and therefore costs ple’s value. Quality is also related to cost since it less. Some data will always require complete enu- links directly to operational possibilities and con- meration, such as the information needed to con- straints. Compliant participants, realistic data col- trol quota allocations or fishing licence limits. lection schedules, good training and equipment, The main problem of data from sample surveys, all take time and cost money. Indeed, the likely quite apart from sample error and accuracy, is that quality of data obtainable from the application of the data population from which they are taken available resources for an envisaged data collec- may not be evenly distributed across the location tion scheme will need to be taken into account as (strata) of the data, whether this is in geographic far back as the derivation of the performance indi- space, time or other dimension. Subdividing a data cator under consideration. If the quality of data population into groups or strata and then ran- from a devised system cannot match the required domly sampling those can reduce the data variabil- accuracy of the indicator then no amount of re- ity to that which represents real differences design of the sampling system will substitute, and 90 Chapter 5 its estimation should be dropped or another choice the added uncertainty of changes in inland hydro- of indicator be made. logic cycles and balances. Nevertheless, short- term predictions can enable early responses to, and 5.1.9 System appraisal modifications of, human interventions (Shepherd and Pope, Chapter 7, this volume). This approach Fisheries information cycles require continuous requires commitment to long-term observations and circular reappraisal. Setting up a data collec- of the state and dynamics of ecosystem compo- tion system and then operating it without continu- nents; in particular to regime shifts and alternative ous feedback and possible revision may waste stable states (Shepherd and Pope, Chapter 8, this resources either on unreliable estimates or on volume). Only through the collection of reliable oversampled variables. One of the key implemen- data, from both observations and experimenta- tation guidelines suggested for the adoption of the tion, can new and realistic ecosystem models be precautionary approach succinctly reflects this developed, which can then contribute to longer- issue: ‘develop methods for optimising the moni- term strategic fishery management plans. Invest- toring system’. Appropriate choice of indicators, ments in better information will not only generate matching requirements to resources, good census better stock assessments but will also improve and sampling design, and the continuing evalua- understanding and management of aquatic tion and revision of these are needed to optimize a ecosystems. monitoring system. In addition, new data sources, particularly from satellites, in situ remote sensing and the wider 5.1.10 Information for new reach of surveys, are improving knowledge of directions in fishery research oceanic conditions such as sea-surface characteris- and management tics, currents, topography and bottom type. These will all assist in the estimation of the biological, Detecting trends and patterns of variation over chemical and physical factors that influence time to provide a better understanding of the aquatic ecosystems and fisheries. New techniques causal relationships among and between physical, are also rapidly developing for the genetic charac- economic and sociocultural factors is key to our terizations of fish populations, which will enable a approaches to fisheries sustainability, and this is better understanding of population structure and crucially dependent on the collection of reliable diversity, population mixing and migration (see data over long periods of time (National Research Ward, Chapter 9, Volume 1). Council 1999). These understandings will lead to Notwithstanding these advances, it is an unfor- higher levels of appreciation about the linkages tunate fact that the best scientific evidence may between human population growth, utilization of not be made available in the best form to policy natural resources, environmental degradation makers and even then it will not necessarily be and climate change. The Large Marine Ecosystem properly used. For example, prior to the formula- (LME) approach recognizes five interacting com- tion of the precautionary approach, the uncer- ponents, comprising ecosystem productivity at all tainty recognized in fishery assessments has often levels, fishery resources and their response to ex- enabled fishers to effectively resist management ploitation, ecosystem and habitat health, socio- measures, or allowed fishery managers to resist economic conditions and governance. calls for increased allocation. Human systems The ecosystem approach to fisheries manage- monitoring is clearly part of an overall approach, ment recognizes that the chaotic behaviour of indeed part of the ecosystem approach, to fishery marine ecosystems may never allow accurate management. Institutional mechanisms are ecosystem, hence fishery, forecasts more than a needed (1) to specify the information required, few years in advance (Acheson 1995). This is proba- including non-formal or traditional knowledge, bly doubly true for freshwater ecosystems, given and then adequately communicate this to policy Gathering Data for Monitoring and Management 91 makers, managers and the public, and (2) to analyse There are three general categories of indicators the responses of communities, both individuals that inform fishery policy: (1) general patterns and and institutions, to economic, environmental and trends such as catch, employment, contributions fisheries factors. to GDP (gross domestic product), (2) changes in the In the current debate about community man- infrastructure and institutions that affect fishery agement or comanagement, it is recognized that management outcomes, and (3) indicators, which the key features of the social framework revolve are usually expressed as indices, against some pre- around mutual commitment, understanding of established reference point, in particular in rela- the problems and methods of addressing them and tion to fishing, such as Maximum Sustainable cooperative/collective action, and that this im- Yield (MSY) or Maximum Economic Yield (MEY) plies open and transparent information communi- (see Chapters 6 and 12, this volume). cation. The sociocultural information domain has The four information domains described above been neglected, but is key to our understanding of are a convenient classification of the general the human dimension of ecosystem stability and groups of data required for integrated fishery analy- sustainability. sis and decision making. Each domain can be bro- ken down into a number of classes and within each class there may be numerous data types available for inclusion within the information cycle, some 5.2 FISHERIES permanently included and some temporary or INFORMATION DATA transient. What follows is a general description of 5.2.1 Policy, performance indicators the main data types for the four domains, including and data variables the classes of indicators generally recognized for the domain. (For detailed description of fishery Fishery performance indicators are needed to as- performance indicators and data variables, see sess the effectiveness of fisheries policies. If the FAO 1999.) policy indicates, for example, the requirement to increase employment, this can be interpreted in a 5.2.2 Fishery and operations number of ways through indicators that describe information domain the incremental growth in numbers employed or the relative growth of the employed in relation to This domain includes all the features of the pri- general population growth, whether community, mary capture and processing of aquatic resources: regional or national. Employment numbers them- what amounts of fish were caught, when and how selves will often be an aggregation of a number of and what operational management information is variables, including at-sea, onshore, support and required? ancillary industry workers. In other cases, policy 1 Catch and discards, production and stock may require a reduction in the number of foreign enhancement fishing vessels or foreign crew against domestic These are the key indicators in relation to fish crew, which has been the case in many countries, stock removals and are thus essential for any un- post-UNCLOS, both developed and developing. derstanding of the nature of a fishery. In some In practice, employment policy in fisheries will cases, the only data variables available may be often be a combination of these requirements and from processed product, which then need to be may need the aggregation of data variables from all converted back to whole fish for the estimation of the relevant minor strata. Therefore, policy needs stock removals. To do this, information must be to be clear and detailed in order for fishery man- gathered on conversion factors. This class of indi- agers to determine what indicators are required, cators includes the additions to stocks from en- how to estimate them and what information to hancement programmes such as restocking or collect. species introductions. 92 Chapter 5

2 Fishing vessels and fishing gear physical and chemical nature of the environment. The physical means of fishery production is also All may require an enormous range of data clearly an essential element of basic fishery infor- variables, and many will require both gathering mation and covers the nature of fishing vessels and data from the fishery and experimental or survey fishing gears. Depending on the fishery, a wide information. range of data variables may be required, from ves- sel length to thickness of net twine (Prado 1991). 1 Stock size One of the key indicators long used as a proxy for 3 Fishing effort and sightings stock size is catch per unit of effort (CPUE), which These indicators relate the deployment and activi- is derived from data gathered from the fishery. This ties of the means of production to the catch itself. is one of many cases where indicators are derived Fishing effort is the information recorded or de- for several reasons, and hence should be consid- rived on the deployment of vessels and gear. The ered as the highest priority. Aspects of stock dy- data variables will be different for different opera- namics that affect understanding of stock size are tions. A good understanding of the operational also included, such as knowledge of stock identifi- characteristics of a fishery is always required be- cation and recruitment. For some species, par- fore considering the choice of variable, for example ticularly the schooling pelagics such as sardines accurate position at the beginning and end of a (Clupeidae) and herring (Clupea spp.), fishery- trawl tow will be important for calculating swept independent experimental surveys may be the area but these may not be necessary for purse seine only accurate way of determining stock size or pole and line fisheries. Sightings are generally information. available from patrol vessels, aircraft or other com- pliant fishing vessels and will also contribute to an 2 Stock structure understanding of fishing effort through informa- The dynamics and nature of stocks in relation to tion on place, length of time and type of fishing op- exploitation is dependent on understanding stock erations, including whether or not this activity structure across the life history of the species for occurred in contravention of effort regulations. all cohorts from larvae to reproducing adults. 4 Offences and prosecutions, and the dissemina- Gathering data on the age, size, sex, maturity and tion of compliance information behaviour of individual fish has been the major Compliance indicators are also important for the focus for this class of indicators. Such data management of fisheries and can be used as key in- offer the means to understand the composition dicators of the response of fishers and others to the and performance of individual cohorts in relation fishery management control measures being ap- to the whole stock, from which management plied. Information on the knowledge of control rules can be derived that prescribe, for example, rules by fishery participants may also be used to the age, size or sex of recruits, or seasonal/spatial determine the nature of offences, which can be pre- limitations such as closed spawning seasons or meditated or unwitting, and feedback to develop grounds. improved compliance measures or to change 3 Community structure prosecution practices. Within the ecosystem, indicators of the position, importance and interactions of all species, from 5.2.3 Biology and environment plankton to top predators, will assist in defining information domain the general community structure. This will be par- ticularly important when fishery policy aims to There are three key classes of indicator related to maintain biodiversity, when fishing operates on the biology of stocks: stock size, stock structure multispecies complexes or when conservation re- and community structure. Together these place quirements demand it such as in relation to marine the stock within the ecosystem, alongside the mammals and birds. Gathering Data for Monitoring and Management 93

4 Environment using a variety of variables, enable the develop- Very large numbers of data variables will con- ment of key sectoral growth indicators, such as per tribute to the two key groups of physical environ- capita fisheries product. mental indicators: oceanography/limnology and 2 Investment, subsidy and management meteorology. As with all other indicators, their All forms of financial inputs are essential for the priority will be dependent on fishery and geo- analysis of sector development, particularly in as- graphic requirements. In some areas, for example, sessing market and competition economics. They dissolved gases and temperatures may remain will include direct and indirect investment in the fairly static, for example over coral reefs, while in fishery sector which should include training, and other areas these may fluctuate widely, as in up- the costs of fisheries management, including re- welling areas. Weather, in particular sea-surface search and data gathering, analysis and decision winds and solar incidence, may have important ef- making, and surveillance and enforcement. fects on species’ life histories. Changes in currents or ambient temperatures may affect reproduction 3 Profitability and survival of larvae, or the rate of primary pro- Related to the above, key performance indicators duction upon which fisheries depend (Myers, associated with the generation of capital/profit are Chapter 6, Volume 1). The gross changes, for exam- required together with the technical information ple, caused by the El Niño Southern Oscillation, on plant and machinery and their depreciation that when the enormous equatorial currents of the constitute the sector’s assets. Pacific are reversed, cause major perturbations in 4 Distribution of food, rent and trade the distribution, survival and growth of many Food balances originating from landings, plus im- marine stocks, from tuna to dolphins. ports, less exports, are important social and eco- nomic indicators since they will enable estimation of the changes to per capita food supply and hence 5.2.4 Economic and financial to the contribution of the fishery sector to nutri- information domain tion and health. In similar fashion, economic rent Four groups of information, most of which are is developed from an appropriate combination of common to any production sector, may be used production, prices and costs. At the national scale for economic and financial indicators, including the performance of the sector in support of the (1) the consequences of production such as price, economy needs to be measured through estima- value and employment; (2) financial support tion of the volume and value of trade and foreign which would include investment, subsidy and exchange balances. management; (3) profitability; and (4) the distri- bution of food, rent and trade. This domain is probably one of the most difficult areas for data 5.2.5 Sociocultural gathering in integrated fishery analysis because it information domain requires divulgence of private enterprise financial This most neglected area of information is proba- data that are usually commercial and confidential. bly also one of the most difficult to investigate as it 1 Price, value and employment deals with the multiplicity of measurable factors The price of fish is one of the key variables, and non-measurable perceptions that make up together with catch and effort, that is used in nu- human society. It includes three general groups merous fishery performance indicators, from the of indicators: the characteristics of relationships development of bioeconomic models and refer- which include access to and dependence on fish- ence points (Maximum Economic Yield), to con- eries and the social status of participants; the de- tributions to GDP or access fees (Hannesson, mographics of fisher, processor, marketing and Chapter 12, this volume). Employment figures, support industries, and their activity patterns; and 94 Chapter 5 distribution of income and food. Some sociocul- fundamental to the strategies for the collection of tural indicators are amenable to the definition of data and the application of control measures. performance indicators but, unfortunately, many 3 Distribution of income and food are not, so that targets and limits cannot be easily Key economic and social indicators of fisheries defined. Particular national or local fisheries situa- development and management concern the ways tions, policies and traditions will affect the way in- in which the income and food resulting from dicators are defined. resource use are distributed. Data on earnings 1 Nature of access, community dependence and by individual, household or company – and the de- the social status of fishers mographics of these – provide basic socioeconomic The ways in which the authorities, producer or- information from which such indicators can be ganizations and communities control access and calculated. In similar fashion, the contribution of conduct their business by the incorporation of in- fishery resources and products to food security and formation and rules into business practices, and by nutrition are also key indicators of progress. An ex- conflict resolution, constitute important knowl- ample is the data required to assess per capita fish edge under which fisheries management operates. consumption and hence diet balance. The nature and extent of dependence on a fishery is also a fundamental factor that drives or con- tributes to fishery management policy, including 5.2.6 Summary of employment, income and consumption, and his- information domains torical and cultural association. For example, As with all forms of sector analysis, there is always these factors are major influences on the European overlap between the data requirements of different Union’s Common Fisheries Policy and regional information domains. For example, the numbers support mechanisms. The financial or cultural of fishers contribute to a variety of indicators, in- values that fishers place on fisheries play an addi- cluding fishing effort and catch per effort, equity tional role in the determination and evaluation of and distribution, and employment and profitabil- fisheries policy, including the likelihood of com- ity. The numbers of fishing vessels contribute in pliance with control measures. similar ways. 2 Participant demographics and activities Each issue, from determining fishery policy to The characteristics of fishers and others con- the means of obtaining the data, needs a full con- tribute to the derivation of fisheries policy and sideration of how the steps in the fisheries infor- management measures, and may be important in mation cycle can be analysed and used in decision defining the structure and stratification of data making. This consideration can be undertaken by collection systems. Fishing and processing prac- addressing each issue against a hierarchy of terms. tices are often seasonal, using a variety of methods. Table 5.1 provides an example of just one eco- Fishing locations and methods for target species, nomic dimension against a hierarchy of issues that involving different sections of the fisher commu- may apply to any dimension in fisheries. nity and their decision-making processes, require an understanding of these patterns. For example, the seasonal migration of nomadic peoples to the 5.3 FISHERIES DATA coast in Oman for subsistence sardine (Stolepho- COLLECTION AND rus sp.) harvesting, and involving all community MANAGEMENT members, overlaps with the near-shore rock lob- ster (Palinurus sp.) harvesting season which is con- Obtaining information from the fisheries produc- ducted by men. These fisheries, while conducted tion sector can be a costly and onerous task, and by the same community, involve completely dif- has often led to adversarial relationships between ferent demographics, and knowledge of these is the sector and fishery authorities. The problem Gathering Data for Monitoring and Management 95

Table 5.1 Hierarchy of terms in fisheries information (4) allocation among users and over time, and (5) (using economics as the policy dimension). compliance control and enforcement. The more involved fishery users are, and the closer agree- Step Example ment between them and fishery authorities, then Policy dimension Economic the more successful will be the conduct of fisheries Objective Economic efficiency management. The development or improvement Criteria Capital productivity of fisheries data collection and management Reference points Limit – capital productivity at bioeconomic schemes thus begins at the strategy level. equilibrium = 0 Target – capital productivity set by policy (e.g. MEY) 5.3.1 Data collection strategies Indicator Financial net return in relation to capitalized value ($return/$capital) Whatever the objective or performance indicator Scale Fishery, fleet required, any strategy will follow a series of natural Variables Income, investment, vessel replacement value, steps: depreciation rate, inflation index • Historical data: evaluation of existing data, its Data types Total landings value and costs, taxes and content, availability and accessibility. subsidies, capitalized value • Operating characteristics: evaluation of the fish- Data sources Fishery administrations, banks, industry, companies ery, fleet, markets, or institutions through census or frame survey. • Approach: decide on complete enumeration or sampling; conduct analyses of cost-benefit and relates partly to a general lack of compatibility cost-effectiveness; assess operational potentials between individual/company/community values and constraints (institutional, financial and and those of the fisheries authority, which sets the human resource requirements). political, legal and administrative frameworks. It • Design: decide on stratification; develop collec- also relates partly to the necessity to impose re- tion methods such as data forms, standards and strictions on fishery rights, hitherto perceived as training. inherited territorial or stock-user rights. Addition- • Test: conduct test or pilot schemes with stake- ally, there is the question of the aims of data holders, such as data collectors, companies, com- collection. Without feedback on why data is munities or other institutions, or a selection of required, users may question its applicability: ‘We these, to validate the design and methods. never see the results of all this effort to collect • Feedback: review results of tests to ensure that data.’ This lack of feedback can lead to distrust, to data types, quantity, quality and origin are suffi- commercially confidential data being leaked to cient and necessary for the estimation of the per- competitors, or data being used for purposes other formance indicator. than the purpose for which it was originally Effectively, this information strategy implies loop- intended, for example for legal or financial in- ing through the entire information cycle as a pre- vestigations. Whatever the causes of conflict or liminary trial prior to full implementation. discontinuity in the information relationship, it seems clear that transparency and responsibility 5.3.2 Stratification in the overall process will tend to remove the obstacles. Two levels of stratification are generally used to Comanagement can help synchronize public- subdivide data sources; major and minor strata and private-sector values through commitments (Table 5.2). Major strata are natural groupings out- of both groups of the fishery sector to the tasks of side the control of the design. Minor strata, which fisheries management: (1) assessment, (2) setting sub-divide the major strata, are selected as part of of objectives, (3) selecting management measures, the design process to reduce data variability within 96 Chapter 5

Table 5.2 Examples of major and minor fishery stratifications.

Major strata Minor strata

Spatial Province of country or major city – Districts (islands, villages) – Home port (place of registration) – Base port of fishing – Community of residence – Landing place – Fishing grounds Time Fishing season – Basic time period (week, month, year) – Day/night Enterprises Companies/cooperatives – Processing plants – Type of support industry Trade Markets/auctions – Intermediaries/companies – Exporters/importers Vessel/gear group Fishing fleet – Gear – Fishery (defined by fleet/target species/gear) – Vessel type (small scale, semi-industrial, industrial, joint venture, foreign) Experimental fishery Geographical areas/depth zones/bottom types/habitats – Time period/day–night – Gear/fishing operation Landings Commercial species group (catch/effort, value) – Commercial size/treatment group (catch/effort, value) – Ecological species groups – Landings agent People or households Demographic subgroups – Fishing community – Fishing fleet – Economic sector (harvest, post-harvest, market, support industry) – Status (captain, crew, owner) Environment Habitats (floodplain, lake, mudflats, mangroves, upwelling areas) – Season – Physical oceanographic/ limnological criteria

the stratum, to produce homogenous subsets and so that reduction in resources leads to a planned minimize statistical bias. reduction in the collection of lower priority data, Choice of minor strata can be complex. To sub- and not to emergency cuts which have unknown divide strata into groups with the highest degree of consequences. For example, if the number of data homogeneity means that data collection will often recorders is reduced by 25%, is the sampling fre- need to sample a large number of strata. However, quency reduced, or are certain data sets omitted increased strata sampling requires more resources. altogether, while maintaining sampling rates on In practice, a balance of available resources and sta- the higher priority variables? Once a scheme is tistical stratification requirements always needs adopted and resourced, it is fairly simple to plan to be made. for reductions, or increases, say, if some limits are temporary. 5.3.3 Operational considerations 5.3.4 Data sources and The financial, human and institutional resources collection methods required to undertake data sampling and complete enumeration can often be considerable. Clearly, The wide range of data required for fisheries man- the choice of performance indicator and the data agement is generated at all levels in the fisheries variables it needs, and the methods of collecting sector; thus sources include harvest, post-harvest, the data, are constrained by the availability of market, support industry, consumer and govern- these resources. Careful estimation of resource ment agency. Unfortunately the simple vertical needs at all stages can be undertaken and tested transmission of such data to fishery management during strategy development, but even during full authorities often ignores the needs of stakeholders implementation of a data collection programme in the management of their own roles. Feedback of there needs to be some flexibility. This should be available information that is of benefit to stake- built into the overall data collection programme, holders always encourages better participation in Gathering Data for Monitoring and Management 97 the provision of data. The management of data full spectrum of general data needs, but it is also a flows, as a key function of a cooperative manage- particularly useful method when seeking to gather ment framework between the fishery sector and perceptions and opinions. governance, forms a major consideration in the • Interviews: Face-to-face interviews are directed development of comanagement arrangements and and compiled by data recorders, often called fish- institutions. ery enumerators. As with questionnaires, the de- There are always information flows between sign of the survey forms and the questions posed the various subsectors. The expected catch from a needs to take into account the likely availability fishery is used to plan private investment in fish- of the information and the capacity of the inter- ing vessels and plant, or bring them into operation. viewee to provide it. This is most often the best The prediction of landings is used to plan process- method for small-scale and artisanal fisheries ing and storage capacity. Quota allocation is used where limited literacy or inadequate communica- to plan marketing and sales. Consumer prefer- tion channels are available. This form of data col- ences are used to plan products (Young and Muir, lection through structured ‘interview’ is also the Chapter 3, this volume). For fisheries comanage- method most employed for sample surveys, where ment to succeed, a vital part of the role of fishery enumerators sample the required data variables authorities is to facilitate these exchanges, in addi- under a specific frequency or location schedule. tion to their more traditional requirements, for an Less structured approaches to interviews include understanding of the biological and economic the use of (1) focus groups, which represent a par- status of fisheries. ticular issue, and (2) panel surveys, in which a ran- As part of the planning of a data collection pro- dom sample of respondents from a group is used as gramme it is valuable to identify the matrix of data a representation of the whole group. types required against the likely primary source, • Direct observations: Obtaining data through di- plus the secondary sources, which can be used to rect observation in ways that do not require input validate the data. Logbooks from fishing vessels – from fishery participants is the most common way as primary sources of operations data – can be veri- of obtaining independently verifiable information. fied using records of landings and these in turn can Observers (at-sea, or at landing sites) record fishery be verified through factory output information. operations and biological characteristics, whereas The methods of data collection are extremely inspectors are deployed to also collect information varied and depend on the nature of the variables. that may be immediately used for compliance con- Five different groups of methods can be employed, trol, such as quota management, the enforcement even within a single fishery: of management measures such as closed areas, or • Registration: This is often the most direct for data verification and validation. Scientific re- means to obtain complete enumeration of the searchers independently observe the biological, means of production and is usually implemented population and environmental characteristics of a through some form of licensing arrangement re- fishery. This can often be supplemented by the use quired by law; it can apply to fishing vessels, facto- of key informants, whose specialized knowledge, ries, fishing gear, companies or individuals. rank or skill enable direct observations beyond • Questionnaires: These are forms that have to the scope offered by the employees of the fishery be completed and submitted by individuals. Care authority, for example academics, community must be taken in their design to require only data leaders and wardens, or cooperating fishers and that are relevant to the respondent. If such ques- companies. Lastly, technology may be employed tionnaires are routine, any supporting documenta- through in situ or remote sensing for data genera- tion – how to complete the form, etc. – might also tion, including the use of satellites and data buoys, include the results of previous questionnaires, or the operation of vessel monitoring systems thus providing feedback. The levels and types of (VMS). The opportunities for VMS are still in their questionnaire information may range across the early stages of development, but it is likely that 98 Chapter 5 increasing numbers of fishing vessels will be sub- With the growth of high-speed telephone lines, ject to these devices in the future. With integrated area networks and the Internet, maintenance of instrumentation it is believed possible to inde- data can be undertaken in a distributed manner pendently verify, along with the Captain and his rather than through the centralized processing of company, most variables deriving from fishing paper or mainframe computer systems. This can operations and from the environment, including ensure that data entry and documentation and climate and oceanography, along with precise primary data processing are undertaken as close as spatial and temporal data on the conduct of a fish- possible to the data source and by those responsi- ery and its location. ble for its collection. • Reporting: VMS can also offer a means to dir- So much of the primary data required for fish- ectly report details of catch and effort and hence eries management is considered as commercially substitute for paper reporting. However, most confidential that it is important to ensure that data fisheries rely on the regular submission of log- management systems protect it from corruption books, which is often a legal condition of licensing. and unauthorized access. Indeed, even within ‘offi- They often provide highly detailed information – a cial’ groupings, data that reveal the identities of complete enumeration of activities – that enable a catch, income or investment may not be necessary range of stock assessment methods, and can con- for their analysis, and therefore access to that level tribute to ecosystem research. At the post-harvest, of detail may be limited. Data control and security sale and trade levels, records kept for commercial features should always form a major consideration purposes can also be required. For example, it may for the human–computer interface, whether this is not be possible to estimate the mass of small pelag- related to the data entry procedure, data access or ics that are landed in bulk until scale readings from to the dissemination/publication of analyses. conveyor belts or pumping systems are taken; or, species composition from an artisanal fishery can- not be assessed until grouped at the market for 5.3.6 Planning and implementation sale. On the financial scale, where required perfor- Planning phase mance indicators include structural and economic assessments, regular reporting through fishery The information cycle described above also pro- statistical surveys, which may be voluntary or vides the basis for planning and implementation of compulsory, may often be required. data collection strategies and programmes. Dur- Again, the development of an information matrix ing the planning phase it is essential to ensure that that identifies the data required with an appropri- fishery policies are supported by appropriate legal ate method for each fishery is a useful tool in devel- mechanisms, including the powers to obtain in- oping a strategy for a data collection programme. formation. Institutional mechanisms are also needed, most often by establishing linkages through 5.3.5 Data management committees and working groups and the proce- dures and communications necessary for infor- Fishery data can often be used for several purposes, mation flow between the private, public and some of which require the raw form, some aggre- non-government sectors. gates. Therefore, all data should be stored in their Working practices need to be established that primary form. This is partly to ensure data integr- enable a data collection programme to proceed. ity and validity, and partly to offer the opportunity Ideally, these will be designed around the tasks, to aggregate or transform data without the loss of rather than the structures, performance or meth- information. ods of the administration. This is not always pos- Computerization and the use of database man- sible, given the nature of bureaucracies, but an agement systems for all forms of data are now the understanding of working practices and their flexi- established norms for information management. bility to respond to change is vital for planning any Gathering Data for Monitoring and Management 99 new or revised programme. This has clear links data contain, through the use of data tables and to the financial and budgetary aspect of planning. data graphics. Once instituted, a data collection programme should develop the time series from which fishery 5.4.1 Data tables performance analyses are made. Medium- to long- term budgetary allocation is essential prior to im- In the first case, the evidence is the data itself, or plementation to ensure the required continuity. aggregates arranged in appropriate strata. Data However, planning and strategy development tables are essential for this proof and typically con- should also account for enforced budgetary re- sist of columns of independent variables (time, ductions, through prioritization, sample scheme space or major strata) with their associated num- flexibility, and a clear description of their bers in the rows described by minor strata or their consequences. combinations. Thus, a time series of catch, by fish- ery, by species would typically be represented as in Table 5.3. Implementation phase Key presentational principles are shown in this Trial or test programmes prior to implementation table: not only verify the methodology, they also test the • Arrangement: strata type and the independent acceptance and level of support from the data sup- variable in columns; major strata (in this case fish- pliers. This support can be maintained during ing method) and aggregate data highlighted; and implementation in many ways, including the minor strata (in this case species data) are sub- demonstration of the benefits and the provision of sidiary to major strata. incentives, or by the application of penalties. • Format: numbers right-aligned; same font as Continuous feedback is required to support text with slightly smaller point size; table margins data collection programmes and to ensure their indented from text and centred on page; optimum continued applicability, as emphasized in the in- use of white space; and frames (grid lines) are mini- formation cycle. Thus, training in new or revised methods as programmes proceed, and consulta- tion to ensure data collection practices are the most up-to-date and meet all stakeholder require- Table 5.3 Total catch in metric tonnes by fishery by ments, should be maintained as an integral part of species, 1996–9 (hypothetical). implementation. Fishery/species 1996 1997 1998 1999

Longline 17200 19300 18300 20700 5.4 FISHERIES INFORMATION skipjack 10000 11000 10000 12000 PRESENTATION albacore 1200 1300 800 700 yellowfin 6000 7000 7500 8000 Fisheries data most often relate to time series or Demersal trawl 26000 24000 23400 31200 spatial distributions (two- and three-dimensional). cod 25000 22500 22500 30000 Sometimes both time and space – the four dimen- hake 1000 1500 900 1200 sions – are required to make sense of the data. In Pelagic trawl 50000 40000 45000 65000 other cases, data sets may consist of two or more mackerel 15000 12000 13500 23000 variables, which may or may not be related to one herring 35000 28000 31500 42000 another. To cope with the often-large data sets that Purse seine 35000 12000 32000 25000 are generated it is usually necessary to present herring 20000 7000 18000 17000 data in ways that (1) provide the evidence upon sardine 15000 5000 14000 8000 which statistical conclusions are made and (2) Total 128200 95300 118700 141900 provide a means to reveal the information that the 100 Chapter 5 mal, clear, of minimum weight, and placed to de- • presenting many small numbers in a small lineate major strata. space, and making large data sets coherent; • encouraging visual comparison of differences 5.4.2 Data graphics and revealing the data at both lower and higher lev- els of detail; and Using simple principles of graphic presentation • serving a clear purpose and being closely inte- usually leads to a better view of the information grated and relevant to the statistical and verbal content of data. Indeed, graphics are often the descriptions of the data. only way of revealing information since they Figure 5.3 contains some of these principles using can be used to accurately depict large data the data from the major strata in Table 5.3. This sets, which are nearly always multivariate. There graphic depicts some design principles that sup- are numerous ways of visually presenting data: port the way information in presented as shown histograms (column, bar), line and scatter above, including: plots, proportional plots (pie, area, radar) to three- • The representation of numbers as physically dimensional surfaces. The most common is the measured on the surface of the graphic should be column histogram, which enables standard pre- directly proportional to the numerical quantities sentation of the independent variable along the represented. (Note: the common use of three- x-axis, for example the passage of time, and a dimensional bars adds nothing to the information variety of ways of presenting data for single and content, and a poor choice of perspective can even aggregated strata. lead to visual distortion.) In general, graphic presentations should reveal • Clear labelling should be used to minimize dis- the information content of data by: tortion and ambiguity about the data. Legends • showing the data by representing the substance should be placed in support of the data. rather than the method of presentation; • Variation in the data should be shown, not de- • avoiding distorting the data – either graphically sign variation. The choice of stacked columns here or by showing the data out of context; provides visual comparison of the variations in

160

140

120

100

80

60 Metric tonnes (thousands) 40

20

0 1996 1997 1998 1999 Fig. 5.3 Total catches (metric Longline Demersal trawlPelagic trawl Purse seine tonnes – thousands) by fishery, 1996–9 (hypothetical). Gathering Data for Monitoring and Management 101 both fishery and total annual catches. (Note: mul- ronment, their ecosystem and the nature and ex- tiple graphics within documents should use the tent of the human interventions on them. Infor- same or very similar formats, as they often also mation uncertainty, unreliability or inadequacy form a visually coherent series.) translates into inability to establish the prudent • Maximize the ink related to the data (areas, foresight that the precautionary approach de- points, lines and text), which draw the eye to the mands. Nevertheless, if proper understanding of data itself. Minimize the ink related to design fish stocks and the appropriate application of fish- (grids, axes, fills and other decoration – often called ery management measures are to be achieved, chartjunk). (Note: the common use of cross- there are principles that can be applied to the col- hatching creates moiré effects that can visually lection and presentation of data which will inform disturb the information content of the graphic.) the knowledge system upon which decisions for Although not presented here, the often-depicted sustainability can be made. times series of money (and derived indicators) The fisheries information cycle begins with should usually use deflated and standardized units the recognition of the areas in the fishery system of monetary measurement rather than nominal over which we can exert some influence. Fisheries units. For a wider appreciation of data graphics, management controls human intervention, the refer to the Graphics Press series by E.R. Tufte fishery, but also links to the prevention of harmful (1987, 1990 and 1997). man-made environmental disturbances. It is dri- ven by policy decisions and management objec- tives. These must be clearly defined both to enable 5.5 CONCLUSIONS managers to implement them, and also to have methods to assess whether decisions have worked, Fisheries management faces a complex web of rela- and what to do about it if they have not. Fishery tionships between the environment, the ecosys- performance indicators are thus defined by the tem and human interventions; each is part of an objectives of policy. Stock sustainability is quite overall ‘natural’ system, which can and should be clearly key to the whole system and may require a managed with the goal of long-term sustainability. range of performance indicators in understanding The sustainability of a fishery, in itself, is impor- how this is to be achieved. Assuming targets can be tant for human welfare reasons, but not at the established for performance indicators (and some, expense of stock (or biodiversity) deteriorations, particularly sociocultural, may be very difficult) which would result in benefit declines. Similarly, the observable data variables can be defined, to- stock sustainability, in itself, is important for gether with a wide range of data characteristics, ecosystem reasons, the consequences of deteriora- from prioritization to standardization. Once data tion potentially having environmental effects, variables are defined (why and what) it becomes a known or unknown, now or in the future. Informa- practical task to collect them in a manner that is tion on the components of this web and the dy- amenable to statistical and other analytical proce- namics of their interactions must take a holistic dures (how). This is usually accomplished through approach; no one area is more relevant than the establishment of a data collection strategy or another, even though the weighting and priority through strategies that take into account various of some data may be higher than others. operational constraints and likelihoods, which Gathering data for resource monitoring and then indicate the methods that need to be applied, fisheries management is an expensive and highly including the temporal and spatial limits that sat- complex task for which there are no single isfy statistical evaluation. Introduction of the data prescriptions. Fish stocks and the fisheries that collection system that implements the methods depend on them are all unique because the will involve a number of actions including the conditions that affect them are always uniquely design of databases to hold the information; legal, characterized by their physical and chemical envi- institutional and budgetary planning; training in, 102 Chapter 5 and monitoring and administration of the system, scientific papers. Technical Consultation on the Pre- including mechanisms for system appraisal that cautionary Approach to Capture Fisheries (including may feed back and influence any level of the infor- Species Introductions), Lysekil, Sweden, 6–13 June 1996. FAO Fisheries Technical Paper, no. 350, Part 2, mation cycle. Rome: FAO. Integrated fisheries information management FAO (1999) Guidelines for the routine collection of systems, through which all information domains capture fishery information. FAO Fisheries Technical are properly represented and their data made avail- Paper, no. 350. Rome: FAO. able for analysis, are the key to the present and Gabriel, W.L. and Mace, P.M. (1998) A review of reference future ability to undertake sustainable fisheries. In points in the context of the precautionary approach. the past, attention has been heavily weighted to In: Proceedings of the Fifth National NMFS Stock Assessment Workshop: Providing Scientific Advice to the biology and population dynamics of fish stocks Implement the Precautionary Approach under the for good reasons; it was the essential start. How- Magnuson–Steven Fishery Conservation and Manage- ever, if fisheries are to be managed at or near the ment Act, 24–6 February 1998. NOAA Technical global limits now becoming recognized, it is cru- Memorandum NMFS-F/SPO-40. cial that all information domains enter into con- Larkin, P.A. (1997) The Costs of Fisheries Management sideration, in particular the human elements that Information and Fisheries Research. Proceedings of the Second World Fisheries Congress. Australia, characterize a fishery, both now and under some CSIRO. future sustainable arrangement. National Research Council (1999) Sustaining Marine Fisheries. Washington D.C.: National Academy Press. Prado, J. (1991) Guide for essential field observations con- REFERENCES cerning fishing gears and vessels in small-scale fish- eries. FAO Fisheries Circular, no. 845. Rome: FAO, Acheson, J.M. (1995) Environmental Protection, Fish- Fisheries Dept. eries Management and the Theory of Chaos: Improv- Tufte, E.R. (1987) The Visual Display of Quantitative ing Interactions between Coastal Science and Policy. Information. Connecticut: Graphics Press. Washington, DC: National Academy Press. Tufte, E.R. (1990) Envisioning Information. Connecti- Caddy, J.F. and Mahon, R. (1995) Reference points for cut: Graphics Press. fisheries management. FAO Fisheries Technical Paper, Tufte, E.R. (1997) Visual Explanations: Images and no. 347. Rome: FAO, Fisheries Dept. Quantities, Evidence and Narrative. Connecticut: FAO (1982) Definition and classification of fishing gear Graphics Press. categories. FAO Fisheries Technical Paper, no. 222. UNCLOS (1982) United Nations Convention on the Law Rome: FAO, Fisheries Dept. of the Sea. United Nations. FAO (1985) Definition and classification of fishery vessel United Nations (1995) Agreement for the Implementa- types. FAO Fisheries Technical Paper, no. 267. Rome: tion of the Provisions of the United Nations Conven- FAO, Fisheries Dept. tion on the Law of the Sea of 10 December 1982 FAO (1995) Code of Conduct for Responsible Fisheries. relating to the Conservation and Management of Rome: FAO, Fisheries Dept. Straddling Fish Stocks and Highly Migratory Fish FAO (1996a) Chronicles of marine fishery landings Stocks. (1950–1994): trend analysis and fisheries potential. World Customs Organisation, United Nations (1994) FAO Fisheries Technical Paper, no. 359. Rome: FAO, International Convention on the Harmonized Com- Fisheries Dept. modity Description and Coding System, Brussels, 14 FAO (1996b) Precautionary approach to fisheries. Part 2: June 1983. Part 2 Stock Assessment

6 Surplus Production Models

JON T. SCHNUTE AND LAURA J. RICHARDS

6.1 INTRODUCTION process analogous to the growth of individual animals. Thus, the collective biomass has an A sustainable fishery requires a productive fish asymptotic capacity similar mathematically to stock. On an evolutionary scale, most stocks have the asymptotic size of a single fish. Our analysis in adapted to a long history of changing circum- this chapter makes this analogy precise through a stances. Theoretically, this implies an ability to differential equation that links historical growth respond to sudden new causes of mortality, such models with the production models of Schaefer as that imposed by a fishery. If the population (1954, 1957), Pella and Tomlinson (1969) and Fox responds with additional productivity to compen- (1970, 1975). Smith (Chapter 4, this volume) pro- sate for fishing mortality, then this increase can be vides an historical perspective on the development called ‘surplus’ production. of surplus production models. A finite carrying capacity represents one pos- The modern approach extends these classical sible mechanism for surplus production. Accord- models by including more biological detail, such ing to this scenario, the biomass expands to a level as explicit assumptions about recruitment, mor- that can be sustained by the environment. If the tality and growth (Sparre and Hart, Chapter 13, biomass becomes larger, then competition for this volume). The emphasis has shifted from dif- scarce resources increases overall mortality and ferential to difference equations that relate more causes the biomass to contract. Fishing imposes obviously to the data available from most fish- mortality that reduces the biomass below carrying eries. All models, whether classical or modern, capacity. At this lower level, reduced competition deal with three essential questions: among surviving animals and greater relative 1 What processes govern the population dynamics abundance of resources allows the population to re- of fished stocks? spond with surplus production. Theoretically, the 2 Given these processes, what is a ‘safe’ harvest biomass can be held indefinitely at a level below level that produces an ‘optimal’ catch while main- carrying capacity, and the fishery can continue to taining a ‘healthy’ stock? operate on the corresponding surplus production. 3 What do the available data reveal about the Essentially, every assessment model of a sus- assumptions in question 1 and the answers to tainable fishery involves the concept of surplus question 2? production, although models vary in their detailed Question 2 requires defining certain key cri- descriptions of population dynamics. Early pro- teria encapsulated by vague adjectives: safe, opti- duction models used differential equations to de- mal and healthy. One classical approach comes scribe a single stock restricted by a finite carrying from the carrying capacity scenario described capacity. These represent population growth by a above. A closed fishery produces no catch, and the 106 Chapter 6 stock maintains itself at carrying capacity. By con- assumptions, which might be wrong. Some of the trast, a very intense fishery drives the population models presented in this chapter have remarkable to extinction, leading eventually to zero catch. elegance, but this does not guarantee that they Theoretically, some intermediate level of fishing correctly represent nature. Given the complexity mortality should give the maximum sustainable of marine ecosystems, a modeller must treat any yield (MSY) under equilibrium conditions. In the stock assessment model with extreme scepticism Schaefer model, for example, the equilibrium rela- directed towards assumptions that might fail. No tionship between yield and biomass has a para- single approach offers a panacea that applies to bolic form, where maximum yield occurs when all situations (Sparre and Hart, Chapter 13, this the biomass is held at 50% of carrying capacity. volume). Thus, the model suggests this level of abundance We focus our discussion on two surplus produc- as a reference point for a healthy stock size associ- tion models that encapsulate classical and modern ated with an optimal catch. Both classical and points of view. The first uses a differential modern models can be used to define a variety of equation proposed by Pella and Tomlinson (1969). similar reference points, depending on goals set for We compare this framework with difference the fishery and the marine ecosystem. equations obtained from a more extensive age- Mathematically, the three questions cited structured analysis (Schnute and Richards 1998). above correspond to three distinct phases of the These two approaches have features in common, analysis. First, a model must be articulated that de- but also significant differences. Each example scribes the hypothetical stock dynamics. Second, gives mathematically tractable results that shed model properties must be investigated to deter- intuitive light on the inner workings of fishery mine the effects of various fishing strategies. Third, models. Our comparison leads to a few results quantities within the model must be compared never published previously, although other as- with actual data to estimate reference points pects of the material appear in greater detail else- needed to implement the chosen strategy. In mod- where (e.g. Quinn and Deriso 1999). ern terminology, quantities internal to the model Consistent with the history of this subject, our arecalled‘states’andthemodelitselfisa‘state space presentation has a high technical content. For per- model’. For example, the model might describe the spective, we organize most of the mathematics dynamics of stock biomass, which is not observed into tables, and focus primarily on the underlying directly. This hidden state variable does, however, concepts. We often omit proofs, but emphasize influence various observed quantities, such as the logical relationships between assumptions and catch or an abundance index. The available data conclusions. We regard models as tools for may or may not be adequate to estimate all the hid- thought, rather than ultimate descriptions of den states, as well as fishery reference points. reality. These tools can be used to the greatest Anyone wishing to pursue the history of pro- advantage by understanding them in detail. Fea- duction models can expect a fairly intense math- tures of simple models can guide the development ematical journey. A reader need only glance at the of more complex models. We ask readers to ap- historical papers by Pella and Tomlinson (1969), proach our material with a sense of adventure. Fox (1970, 1975) or Schnute (1977) to find equa- Scaling a mountain to gain perspective requires a tions in abundance. Mathematical models serve as passage through numerous difficulties and lesser metaphors of reality. The strength of mathematics vistas along the route. lies in its ability to explore consequences. A few Equations in our summary tables follow a sim- assumptions can produce an elaborate description ple numbering system; for example, (T6.2.3) refers of fish population dynamics, where mathematical to equation 3 in Table 6.2. Subscripts usually indi- results guide biological interpretation and under- cate time or age variables, which may be discrete standing. On the other hand, mathematics is or continuous, depending on the context. For in- limited by its complete dependence on the stance, Bt, wa, and w• denote biomass at time t, Surplus Production Models 107

5 100

4 + 80 ) ) C

3 60 B Fig. 6.1 Key concepts in surplus production models, where the

2 40 Biomass ( equilibrium catch rate C (solid Catch rate ( line) and biomass B (broken line) depend on the fishing mortality 1 20 rate F. Symbols indicate the following conditions: pristine stock (᭺); fishery at MSY (᭛); 0 0 stock extinction (ᮀ); and a 0.0 0.02 0.04 0.06 0.08 0.10 0.12 precautionary fishery with catch rate lower than MSY (+). Fishing mortality rate (F)

ۋ weight at age a, and asymptotic weight as a Æ•. Et Bt, Ct. (6.1) An asterisk indicates a value associated with sta- ble MSY conditions. Thus, the maximum sustain- In this scenario, the biomass Bt usually acts as a able catch rate C* occurs when the biomass B* hidden state that cannot be observed directly. experiences an optimal fishing mortality rate F*. A Thus, in practical terms, (6.1) reduces to prime symbol denotes a value associated with pris- ۋ tine conditions, that is, a stable unfished stock. Et Ct, (6.2) (Remember that ‘prime’ means ‘pristine’.) Thus, B¢ is the pristine stock biomass, sometimes desig- which translates to the more romantic statement nated B0 in other fishery literature. We retain the ‘Boats go out to sea and come back with fish.’ temporal meaning of the subscript 0, so that B0 de- notes the initial biomass at time t = 0. This may or may not be the pristine biomass; thus, the condi- 6.2 GRAPHICAL MODEL tion B0 = B¢ indicates that the model begins with an unfished biomass. Models describe the evolution Before developing formal mathematics, we use of time-dependent states, given a vector Q of model a simple graph to illustrate the key features of a parameters that typically remain constant in time. surplus production model. Figure 6.1 portrays the Sometimes, however, the distinction between long-term impact of a fishing mortality rate F on a parameters and states weakens, and it becomes stock biomass B. In the absence of fishing (F = 0), simpler to think of every quantity in the model as a the stock maintains itself at the pristine level state. B¢=100 biomass units (e.g. tonnes), which is the We give models two levels of interpretation. natural carrying capacity. As F increases, B drops First, we emphasize the biological meaning of as- below B¢, and this reduced biomass leaves room for sumptions and corresponding results. Second, we growth. The fishery harvests the resulting surplus examine each model from a systems perspective, production, generates a catch rate C, and main- in which some states act as controls that deter- tains the biomass at level B. All quantities in Fig. mine other states. For example, an effort rate Et 6.1 represent long-term conditions. Thus, the mor- might alter the population biomass Bt and produce tality rate F maintains the catch rate C by remov- a catch rate Ct, as indicated by ing exactly the surplus production available from 108 Chapter 6 the biomass B. Conceptually, F represents the frac- tion of the biomass: F = C/B. By comparison with tion of B removed per unit time to produce the equation (6.3), the additional parameter catch rate C = FB, measured as the biomass cap- tured per unit time.* C* 5ty-1 y= ==5%y-1 (6.4) A high mortality rate can potentially drive the B¢ 100t population to extinction. In Fig. 6.1, for example, B = 0 when F = 0.13; thus the stock cannot main- gives an overall measure of stock productivity by tain a removal rate greater than 13% per unit time. relating the maximum sustainable catch rate to Following Schnute and Richards (1998), we denote the pristine biomass. this critical point as F†. (Remember that ‘dagger’ Figure 6.1 makes a useful topic of conversation means ‘death’ to the population, i.e. extinction.) with fishermen embarking on a new fishery. Ac- Thus, two conditions imply a catch rate C = 0: no cording to this scenario, the biomass can initially fishing (F = 0) or extinction fishing (F ≥ F†). A mor- be reduced from 100t to the optimal B* = 57.8t, tality rate F between these extremes (0 < F < F†) pro- producing a windfall catch of 42.2t. This might duces a positive catch rate C > 0. In particular, the happen over two years in a fishery that takes maximum sustainable catch rate C* corresponds more than 21t per year. Unfortunately, such a to an appropriate mortality rate F*. The following high mortality rate, much greater than F†, would short table summarizes the three stock conditions rapidly drive the stock to extinction. After identified so far, as designated by accents in the the initial windfall, the catch must be curtailed to text and symbols in Fig. 6.1: a much slower rate near 5t per year. Fishermen who have become dependent on four times this rate may suffer economic ruin. Furthermore, Condition Accent Symbol the figure represents a population that declines F Pristine ¢ ᭺ more rapidly as the mortality approaches the † MSY * ᭛ critical F . In the uncertain world of fishery man- Extinction † ᮀ agement, it might be dangerous to adopt even the theoretically optimal catch rate C* = 5ty-1. Suppose that time is measured in years (y) and A more precautionary approach (marked ‘+’ biomass in tonnes (t). Then Fig. 6.1 implies the in Fig. 6.1) could be achieved with B = 78t and F 5.1% y-1, producing an approximate catch rate following reference points: B¢=100t, F* = = C 4ty-1. Such a policy would reduce the risk of 8.7%y-1, B* = 57.8t, C* = 5ty-1, and F† = 13%y-1. = In particular, extinction by maintaining the biomass at a higher, more robust level. C* 5ty-1 The graphical model in Fig. 6.1 has two compo- F* == =87.%y-1 . (6.3) nents, a dome-shaped catch curve and a descend- B*.57 8t ing biomass curve. A mathematical model that More generally in Fig. 6.1, the fishing mortality represents these components must include at least rate F expresses the annual catch rate C as a frac- three parameters: a maximum level for the bio- mass (B¢), a scale parameter for the catch relative to the biomass (y=C*/B¢), and a location parameter that determines the position of the dome (F*). For example, a relatively high F* in Fig. 6.1 gives a *In some non-equilibrium contexts, this relationship dome shifted towards the right end of the catch between the instantaneous rates C and F implies an accu- curve. Any equivalent set of three parameters can mulated yield Y = (1 - e-Ft)B during the time interval t. Models presented later in the paper clarify such alterna- be used as surrogates for (B¢, y, F*). For example, tive formulations. For now, experienced readers can take our first model includes B¢ explicitly, but captures note of the limit C = limtÆ0 Y/t=FB. y and F* through two other parameters labelled a Surplus Production Models 109 and g. Mathematical analysis of the model gives Table 6.1 Production model 1, based on a differential formulas for (y, F*) in terms of (a, g). Keen readers equation proposed by Pella and Tomlinson (1969). can sneak a glance ahead to equations (T6.3.2) and Model (T6.3.7) in Table 6.3. Furthermore, for those who Q=(B¢, a, g, q , q ) (T6.1.1) want an even better preview of things to come, 1 2 F = q E (T6.1.2) equations (6.12) and (T6.1.6) define the curves C(F) t 1 t g dB aBBÈ ˘ and B(F) portrayed in Fig. 6.1. How do the simple tt=-1 Ê tˆ - FB (T6.1.3) Í Ë ¯ ˙ tt ideas in the graphical model suggest such prolific dt g ÎÍ B¢ ˚˙ mathematics? Read on. dYt Ct ==FBtt (T6.1.4) dt

It = q2Bt (T6.1.5) 6.3 CLASSICAL Second formulation DIFFERENTIAL EQUATIONS 1 g Ê ag- Fˆ BF• ()= B¢ (T6.1.6) (MODEL 1) Ë a ¯

g dBt ()ag- FBttÈ Bt ˘ The scenario in Fig. 6.1 represents equilibrium = 1- Ê ˆ (T6.1.7) Í Ë ¯ ˙ conditions achieved after a sustained fishing mor- dt g ÎÍ BF• ()t ˚˙ tality rate F. Because actual fisheries take place on Solution for time step [t, t +t] with constant Ft populations in flux, an appropriate mathematical -1 g -g model must also capture the system dynamics. ÔÏ a È a Bt ˘ Ô¸ BB= ¢ - - Ê ˆ e--()agFt t (T6.1.8) Theoretically, a stock with a carrying capacity cor- t+t Ì Í Ë ¯ ˙ ˝ ÓÔag- FFttÎÍag- B¢ ˚˙ ˛Ô responds to a system in which the biomass B tends t -1 g -g to move towards a particular state B¢. Differential Ï È B ˘ ¸ Ô Ê t ˆ --()agFt tÔ BBFtt+•t = ()11-- e (T6.1.9) equations can easily be formulated to capture this Ì Í Ë ¯ ˙ ˝ ÓÔ ÎÍ BF• ()t ˚˙ ˛Ô behaviour. For example, suppose that the biomass

Bt follows the differential equation dBt/dt = f(Bt), where state variables. From a systems perspective, ÏfB()><¢0 if B B the reader can confirm that each equation Ì . (6.5) Óf()BBB<>¢0 if (T6.1.2)–(T6.1.5) in the table plays the correspond- ing role shown in the figure.

Then Bt increases or decreases, respectively, if Bt Biologically, (T6.1.2) assumes that the rates of lies below or above B¢. Either way, Bt moves to- effort and fishing mortality are proportional, with wards B¢, which acts as a stable equilibrium point proportionality constant q1. The differential equa- that can be interpreted as the carrying capacity. tion (T6.1.3) represents a variation of that proposed Production model 1, defined by equations by Pella and Tomlinson (1969), where our version (T6.1.2)–(T6.1.5) in Table 6.1, includes a parameter uses parameters more appropriate for the biologi-

B¢ that plays this role. Given an effort rate Et, the cal description here. The right side of (T6.1.3) has model sequentially defines five states: the fishing the properties (6.5) when Ft = 0. Consequently, B¢ mortality rate Ft, biomass Bt, catch rate Ct, cumu- corresponds to the stock’s carrying capacity in the lative yield Yt, and abundance index It. In symbols, absence of fishing. When Ft > 0, (T6.1.3) includes a biomass loss proportional to the fishing mortality ۋ Et Ft, Bt, Ct, Yt, It. (6.6) rate (-FtBt). Assumption (T6.1.4) actually contains two equations. The first defines the catch rate Ct as Figure 6.2 represents model 1 as a flow diagram, in the derivative of cumulative yield Yt, and the sec- which model equations define linkages among ond equates this rate to the loss of biomass due to 110 Chapter 6

might be different from U . Such an index might Effort rate Et t come from a fishery independent survey or even from alternative compilations of catch and effort T6.1.2 data. For example, a CPUE index might be ob- tained from the median catch to effort ratio taken from many individual tows (Richards and Schnute

Fishing mortality rate Ft 1992). This index would not, in general, be the same as the ratio of total catch to total effort from all tows combined. Dynamics Biomass B T6.1.4 A model based on differential equations neces- T6.1.3 t sarily involves instantaneous quantities for the continuous time variable t. The dimensional units of each state variable and parameter provide a use- T6.1.5 ful guide to model interpretation. For example, if effort is measured in boat-days and t is measured in

days, then the effort rate Et would have the units of Index It Catch rate Ct boat-days per day, or simply boats. Conceptually, Yield Yt Et denotes the quantity of gear operating at time t. In (T6.1.3), F represents the fraction of the biomass Fig. 6.2 Flow diagram for model 1 in Table 6.1, showing t B removed by the fishery per unit time. Thus, F the state variables calculated from each equation t t -1 (T6.1.2)–(T6.1.5). must have the units day so that the term FtBt agrees dimensionally with the derivative dBt/dt. From this, it follows that q1 in (T6.1.2) must have -1 units (boat-day) . In effect, q1 gives the fraction of fishing in (T6.1.3). Finally, (T6.1.5) assumes that biomass removed per boat-day of fishing. the index It is proportional to the biomass, with a Similarly, Ct in (T6.1.4) corresponds to the rate second proportionality constant q2. of biomass removal measured, for example, in kg The vector Q in (T6.1.1) includes five param- per day. Rates cancel in the ratio (6.7), so that Ut eters: the carrying capacity B¢, two additional has the conventional units of kg per boat-day. We parameters (a, g) for the differential equation include the cumulative yield Yt to represent the (T6.1.3), and the two proportionality constants (q1, catch removed over a time period. Thus, from q2). We assume that a>0, where this parameter (T6.1.4) the yield during the time interval [t, t +t] is governs the biomass growth rate particularly at small stock sizes (B ª 0). As illustrated below, g de- t+t Yt()+ tt- Yt()== Ctt Cdt, (6.8) termines the overall shape of the catch curve, Út where g>-1. Almost every equation in this chapter has a valid limit as gÆ0, although we generally do where the integral can also be used to calculate the not present the explicit analytical form. mean catch rate Ct. Combining (T6.1.2) and (T6.1.4) gives the Analytical properties of the differential equa- proportionality tion (T6.1.3) reveal important features for its biological interpretation. With a fixed fishing mor-

Ct tality F, the asymptotic biomass B•(F) can be com- Ut ==qB1 t , (6.7) Et puted explicitly as (T6.1.6). As F increases from 0,

the formula shows that B•(F) declines steadily where the catch per unit effort (CPUE) Ut acts as an from the carrying capacity B¢=B•(0). For g>0, the index of the biomass. Model 1 extends this concept population is driven to extinction when F = F† = in (T6.1.5) by allowing another index It, which a/g; thus, B•(a/g) = 0. If g£0, the population can Surplus Production Models 111 theoretically sustain arbitrarily high harvest rates, data only become meaningful when accumulated and B•(F) represents a steadily declining function across time to give, say, the total annual catch from for all F > 0. the yield calculation (6.8) with t=1yr. Thus, appli-

Biologically, B•(F) represents a carrying capac- cation of model 1 depends on a method of integra- ity that is adjusted downward from B¢ by the fish- tion to a meaningful time-scale for the data. For ing mortality F. The original differential equation example, if Ft remains constant during the time (T6.1.3) can be proved equivalent to the second for- interval [t, t +t], then (6.8) and (T6.1.4) imply that mulation (T6.1.7). Compare (T6.1.3) when Ft = 0 t+t with (T6.1.7) when Ft > 0 to see that fishing pres- Yt()+ tt- Yt()== Ctt F Bdt t = t FB tt. (6.9) Út sure effectively replaces a by a-gFt and B¢ by B•(Ft). Equation (T6.1.3) also has the convenient fea- ture that it can be solved analytically if Ft is held Thus, the cumulative yield and mean catch over constant. For example, consider a time step [t, t +t] this interval depend directly on the mean biomass during which the fishing mortality is held at its ini- Bt. Unfortunately, the integral of Bt in (6.9) cannot tial value Ft. If the biomass starts at Bt, then the be computed analytically from (T6.1.8) or (T6.1.9) final value Bt+t is given by (T6.1.8). The slightly for general values of g. more intuitive solution (T6.1.9) comes from the Table 6.2 shows model results for three particu- second formulation (T6.1.7). Biologically, (T6.1.9) lar choices of g, where Bt in (6.9) can be computed says that Bt+t represents a movement from the ini- explicitly if g=-1 or 1. We also show limiting val- tial state Bt towards the asymptotic value B•(Ft). ues of the analysis as gÆ0, although Bt cannot be As we mentioned in the introduction, working represented analytically in this case. Historically, through these details can be a rather intense math- Schaefer (1957) and Fox (1970) used models corre- ematical journey. Conceptually, however, model sponding to g=1 and g=0, respectively. We show (T6.1.2)–(T6.1.5) has just a few key properties that below that the case g=-1 has a strong connection concern us. First, it describes a fishery in which to the von Bertalanffy growth model, although it the effort influences stock abundance, catch, and has anomalous properties as a fishery model. other state variables, as represented by (6.6) and Jobling (Chapter 5, Volume 1) discusses the appli- Fig. 6.2. Second, it captures the concept of an inher- cation of the von Bertalanffy model to the growth ent carrying capacity B¢. Third, it has convenient of individual fish. analytical properties that enable us to explore its Formulas in Table 6.2 have somewhat more global behaviour, such as a reduced carrying capac- obvious biological interpretations than their ity under fishing pressure. Fourth, it represents a counterparts in Table 6.1. For example, (T6.2.1) fascinating piece of intellectual history. At the and (T6.2.4) show clearly that B•(F) Æ 0 as F Æ• time these methods were developed, differential when g=-1 or 0; however, when g=1, (T6.2.6) equations played a common role in population implies population extinction at the critical fish- analysis. Computers were not as readily available ing mortality level F† =a. The results (T6.2.2), then as now, and analytical results greatly assisted (T6.2.5) and (T6.2.7) illustrate that Bt+t Æ B•(Ft) as the understanding of model behaviour. In fact, the time step tÆ•. The formulas (T6.2.3) and exact solutions always serve as valuable tools for (T6.2.8) for mean biomass Bt show a tendency thought. towards B•(Ft), with a correction related to the change from Bt to Bt+t. As mentioned earlier, mean values Bt relate 6.4 SPECIAL CASES more directly to fishery data than instantaneous

(MODEL 1) values Bt. Schnute (1977) investigated this prob- lem when g=1 and arrived at a formula similar to Fisheries do not generally produce useful instanta- (T6.2.9). Consider a fishery with regular time steps neous data, such as the catch rate Ct. Typically, the of length t (e.g. t=1yr). Suppose that the effort rate 112 Chapter 6

Table 6.2 Three special cases of model 1 in Table 6.1, with constant fishing mortality Ft during the time interval [t, t +t].

Von Bertalanffy: g =-1 -1 Ê F ˆ BF• ()=+1 B¢ (T6.2.1) Ë a ¯ -(a+Ft )t Bt+t = B•(Ft) + [Bt - B•(Ft)]e (T6.2.2)

È BBtt+t - ˘ BBFtt= • ()1- (T6.2.3) ÎÍ atB¢ ˚˙

Gompertz: g = 0 -F/a B•(F ) = e B¢ (T6.2.4) e-at B BBFÈ t ˘ (T6.2.5) tt+•t = ()Í ˙ Î BF• t ˚ Logistic: g = 1 Ê F ˆ BF• ()=-1 B¢ (T6.2.6) Ë a ¯

Bt BBFtt+•t = () (T6.2.7) --()atFt BBFBettt+[]• ()-

BB¢ Ê t+t ˆ BBFtt= • ()- log (T6.2.8) at Ë Bt ¯

Ê atBt+t ˆ exp -1 ()at-Ft +t Ë B¢ ¯ a - Ft e -1 ÈÊ aBt ˆ ˘ = exp a --Ft t (T6.2.9) ()at-Ft Í ˙ atBt a - Ft e -1 ÎË B¢ ¯ ˚ expÊ ˆ -1 +t Ë B¢ ¯

Et remains constant during the interval starting at function of Bt. This implies a similar expression time t. Then (T6.1.2) and (6.7) imply that for Bt+t as a function of Bt+t. Substitute these com- puted values of Bt and Bt+t into the right side of (T6.2.8). The result, which involves only means B Ft = q1Et = q1 Et, (6.10) t and Bt+t, reduces to (T6.2.9). Ct 1 Bt ==Ut. (6.11) qE1 t q1 6.5 EQUILIBRIUM (MODEL 1) Substituting (6.10)–(6.11) into (T6.2.9) gives an equation that involves only the fixed time step t, Theoretically, the equilibrium version of a dynamic model gives projections for long-term three parameters (B¢, a, q1), and the observed mean yield in relation to a sustained harvest rate. In data (Et,Ut). Thus, when g=1, model 1 can be con- verted to a difference equation that relates directly model 1, a constant effort rate E corresponds to a to data collected from the fishery. Keen readers can constant fishing mortality rate F = q1E. This level glance ahead to equation (6.24). of fishing pressure eventually gives a stable For the mathematical adventurer, we briefly biomass B = B•(F) and produces the catch rate sketch the proof of (T6.2.9). Eliminate Bt+t from equations (T6.2.7) and (T6.2.8) to obtain a single gF 1 ggaBBÈ ˘ CF=-Ê11ˆ B¢= -Ê ˆ . (6.12) Ë ¯ Í Ë ¯ ˙ equation that relates Bt and Bt. Solve for Bt as a a g Î B¢ ˚ Surplus Production Models 113

Thus, C can be computed from either equilibrium shaped catch curve C(F). Similarly, (T6.1.6) gives value F or B. As can be seen from (T6.1.6), the first the descending biomass curve B(F), where we now equality in (6.12) follows from the equilibrium use the notation B•(F) to denote the asymptotic version of (T6.1.4): C = FB = FB•(F). The second biomass associated with a sustained fishing mor- equality comes directly from (T6.1.3)–(T6.1.4) at tality F. Table 6.3 lists analytical expressions for equilibrium. various reference points in terms of the model This equilibrium analysis brings us full circle parameters (T6.1.1). These results suggest poten- back to the graphical model in Fig. 6.1. As indi- tial alternative model parameters. For example, cated in the preview, (6.12) defines the dome- (T6.3.7) implies that

agy=+()1 ()1+gg ; (6.13)

Table 6.3 Various reference points for model 1 in Table 6.1, expressed as functions of model parameters (T6.1.1) consequently, y=C*/B¢ can replace a in (T6.1.1). with g>-1. We have already seen in the discussion of Fig. 6.1 that an equilibrium analysis requires at least three a E* = (T6.3.1) parameters (B¢, y, F*). Remarkably, equivalent q1()1+ g parameters (B¢, y, g) govern the full population a F* = (T6.3.2) dynamics in the differential equation (T6.1.3) of 1+ g model 1, where Ïa † Ô if g > 0 F = Ì g (T6.3.3) ÓÔ• if g ≤ 0 F*.=+()1 gy1 g (6.14) B* = (1 +g)-1/gB¢ (T6.3.4) C* =a(1 +g)-(1+g)/gB¢ (T6.3.5) The proof of (6.14) follows from (T6.3.2) and (6.13). I* = q (1 +g)-1/gB¢ (T6.3.6) 2 Figure 6.3 illustrates the roles of y and g in C* -+()1 gg yag= =+()1 (T6.3.7) model 1. Curves in Fig. 6.3(a) portray the equi- B¢ librium catch C(F) for various values g, where the

(a) (b) 1.0 1.0

0.8 0.8

0.6 0.6

C/C* 0.4 C/C* 0.4

0.2 γ 0.2 γ

0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 H B/B'

Fig. 6.3 Relative catch rate C/C* for the equilibrium version of model 1 as a function of: (panel (a)) harvest rate H in (6.15); (panel (b)) relative biomass B/B¢. Panels show catch curves for various values of g: (a) -0.7, -0.4, 0, 1, 5; (b) -0.9, -0.5, 1, 5, 20. Arrows indicate the direction of increasing g. Heavy solid and broken curves indicate special cases g=1 and g=0, respectively. Panel (a) assumes that ty = tC*/B¢=0.1, although panel (b) is independent of y. 114 Chapter 6 vertical axis is scaled relative to C* and the hori- below the pristine biomass. For example, in the zontal axis uses the scaling Schaefer case (g=1), a new fishery theoretically could begin by taking half the available biomass H = 1 - e-Ft. (6.15) (B¢-B* = B¢/2). After this, however, the stock’s harvest potential would restrict the annual catch Conceptually, H represents a harvest removal frac- to tC* =tyB¢, usually much less than B¢/2 when tion during the time interval t in the absence of t=1yr. As mentioned in the discussion of Fig. stock production. Each curve in Fig. 6.3(a) has 6.1, fishermen might buy gear and enter the fishery slope (ty)-1 at the origin. We have assigned this at a time when fish are abundant, with little slope the value 10 (e.g. y=0.1yr-1 with t=1yr). A evidence of a low sustainable harvest potential. lower slope (i.e. higher ty) would shift curve Excessive investment seems almost inevitable modes further to the right. Figure 6.3(b) similarly without strong measures to detect and implement represents equilibrium curves C(B) in (6.12) for realistic limits during the early phases of fishery various g. Vertical and horizontal axes are scaled development. relative to C* and B¢, respectively. The general public, with memories that ‘you Biologically, the dimensionless quantity ty rep- could once walk across the water on the backs of resents the fraction of pristine biomass that can be the fish’, may have difficulty with the carrying ca- caught during time t under MSY conditions. This pacity argument. For example, why would a bank would be large for stocks with a high potential for account produce a higher interest rate just because harvest, corresponding to a small slope (ty)-1 at the it contains less money? The sceptical modeller origin in Fig. 6.3(a). Similarly, low values g corre- does well to examine alternative explanations spond to highly productive stocks, for which the carefully. Many things can go wrong with model 1. maximum sustainable catch rate is taken at a high Environmental influences may cause the carrying harvest rate (panel (a)) on a biomass that is small capacity B¢ to change randomly or systematically relative to B¢ (panel (b)). Large values g indicate in time. If so, equilibrium analyses could become weakly productive stocks, where MSY is achieved meaningless. Furthermore, real populations could with a low harvest rate on a biomass near B¢. The be subject to depensation, in which the population case g=1 represents an intermediate scenario, in becomes extinct after dropping below some which MSY corresponds to a biomass B* that is threshold biomass level (Myers, Chapter 6, Vol- half the pristine biomass B¢ (panel (b), heavy line). ume 1). The differential equation (T6.1.3) does not Figure 6.3(b) suggests two anomalous extremes. allow this possibility, because the growth rate From (T6.3.2) and (T6.3.4), the case g=-1 theoreti- dB/dt increases steadily as B Æ 0. cally corresponds to a stock with F* =•and B* = 0. Geometrically, as gÆ-1, curves in Fig. 6.3(b) shift left until the right-hand limb becomes a straight 6.6 GROWTH MODELS line from the upper left to the lower right. Simi- larly, as gÆ•, the curves shift right until the left- Fish growth models provide a conceptual link be- hand limb becomes a straight line from the lower tween model 1 and an extended framework that left to the upper right. In this case, (6.14) and includes more biological detail in the population (T6.3.4) imply that F* =y=C*/B¢ and B* = B¢. dynamics. Table 6.4 shows the familiar von Berta-

Most of the results derived for model 1 follow lanffy growth model (T6.4.1) for weight wa at age a from the two key assumptions (T6.1.3)–(T6.1.4), (see also Jobling, Chapter 5, Volume 1). The curve where biomass dynamics respond to an intrinsic extrapolates forward to an asymptotic size w• as carrying capacity B¢ and an external fishing mor- a Æ•and backward to size 0 at theoretical age a0. tality F. Mathematical logic creates a fascinating This model implies that successive weights at biological metaphor. According to this theory, a regular age intervals t, e.g. t=1yr, lie on a straight healthy fishery typically operates on a stock well line (T6.4.2) with intercept l and slope k. Math- Surplus Production Models 115 ematically, (T6.4.2) can be considered a first-order when g=-1. Furthermore, the extended model difference equation. Given an initial weight wr at satisfies the differential equation (T6.5.4) and the recruitment age r, this equation has the analytical difference equations (T6.5.5)–(T6.5.6). Equations solution (T6.4.3). Thus, model (T6.4.1) with in Table 6.5, except those involving l, have math- parameters Q=(w•, K, a0) is equivalent to (T6.4.3) ematical limits as gÆ0. with parameters Q=(wr, k, l), where (T6.4.4) This quick tour through growth models, using shows the parametric relationships. somewhat unconventional notation, allows us to Table 6.5 extends the growth law in Table 6.4 by portray model 1 in a new light. Formally, the bio- adding a transformation parameter g. Equations mass dynamic equation (T6.1.7) and the growth (T6.5.1)–(T6.5.3) show three equivalent formula- differential equation (T6.5.4) are identical, given tions, where (T6.5.7) illustrates the parametric the notational changes relationships. In particular, the extended model re- duces to the von Bertalanffy model in Table 6.4 taBÆÆ,,ta w BFw••() tÆ-Æ ,a g F t K .

Similarly, the solution (T6.1.9) transforms into the Table 6.4 Two formulations of a simple growth model. growth model (T6.5.2), given the changes = w K a Von Bertalanffy: Q ( •, , 0) -K(a-a0) trÆÆ-+ÆÆ,,,,tt art aBw wa = w•[1 - e ] (T6.4.1) BF••()ttÆ-Æ w,.ag F K Brody: Q = (wr , k, l) with given t

wa =l+kwa (T6.4.2) +t Thus, as mentioned in the introduction, a precise Ï l Ê l ˆ ()ar- t +-wr kk, π 1 analogy exists between population growth to- Ô11-k Ë -k ¯ wa = Ì (T6.4.3) wards carrying capacity and individual growth to Ô ar- wr + lk, = 1 Ó t an asymptotic size. Model 1 suddenly appears as an old friend, seen elsewhere in a different disguise. Parametric relationships For this reason, we associated particular choices of w = w [1 - e-K(r-a0)], k=e-Kt, l=(1 - e-Kt)w (T6.4.4) r • • g in Table 6.2 with names taken from the literature

Table 6.5 Three formulations of a growth model that generalizes the model in Table 6.4.

Model -K(a-a0) -1/g wa = w•[1 +ge ] ; Q=(w•, K, a0, g) (T6.5.1) -1 g -g ÔÏ È Ê wr ˆ ˘ --Ka() r Ô¸ wwa =--11 e ; Q=(w , w , K, g) (T6.5.2) • Ì Í Ë ¯ ˙ ˝ r • ÓÔ ÎÍ w• ˚˙ ˛Ô

-1 g È l Ê -gtl ˆ ()ar- ˘ wwar= +- k ; Q=(w , k, l, g) (T6.5.3) ÎÍ11-k Ë -k ¯ ˚˙ r Differential and difference equations g dw KwÈ w ˘ aa=-1 Ê aˆ (T6.5.4) Í Ë ¯ ˙ da g ÎÍ w• ˚˙ -g -g -g wa+t = (1 -k)w• +kwa (T6.5.5) -g -g wa+t =l+kwa (T6.5.6) Parametric relationships -K(r-a0) -1/g -Kt -Kt -g wr = w•[1 +ge ] , k=e , l=(1 - e )w• (T6.5.7) 116 Chapter 6 on growth models. We included the anomalous The flow diagram for model 2 (Fig. 6.4) shows that case g=-1 for the connection with von Bertalanffy sequential calculation starts in (T6.6.5) and pro- growth. ceeds through dynamic equations that sometimes The Brody line (T6.4.2) lies behind the include time lags. Model 2 uses a discrete time step

Ford–Walford method of estimating w• and k of length t=1. Thus, time indices take only integer (Jobling, Chapter 5, Volume 1). It also plays a sig- values, and the notation t can be dropped. For sim- nificant role in formulating our next model. This plicity, we assume that time is measured in years, result takes a similar form (T6.5.6) in the extended although different time-scales may be appropriate growth model, where -g specifies an exponential for some species. Unlike model 1, states in model 2 transformation to linearize the relationship be- always denote annual values. For example, Ct now tween wa and wa+t. refers to total catch in year t, rather than catch rate at continuous time t. Biologically, model 2 assumes a particular func- 6.7 MODERN DIFFERENCE tional relationship (T6.6.2) between spawning bio-

EQUATIONS (MODEL 2) mass St-r and recruitment biomass Rt. A time lag of r years is presumed fixed and known for the Production model 2, defined in Table 6.6 by species. The model assumes a growth curve

(T6.6.2)–(T6.6.9), expresses a more complex bio- (T6.4.3) with parameters (wr, l, k), where wr de- mass dynamics than model 1. Mathematically, notes the weight of fish recruited at age r. The para- these eight equations sequentially update eight meters l and k appear in the biomass update state variables: recruitment biomass Rt, total equation (T6.6.3), where the biomass Bt at the start biomass Bt, mean weight Wt, fishing mortality Ft, of year t includes newly recruited biomass Rt plus harvest fraction Ht, catch biomass Ct, spawning biomass surviving from the previous year. The lat- biomass St, and biomass index It. From a systems ter component comes from the spawning biomass perspective, fishing effort Et acts as a control: St-1 adjusted by two factors, a survival coefficient 1 -dand a growth factor ۋ Et Ft, Ht, Ct, St, It, Rt, Bt, Wt. (6.16) lk+ Wt -1 . (6.17) Wt-1

Table 6.6 Production model 2, based on difference Intuitively, (6.17) comes from a population of fish equations derived by Schnute and Richards (1998) from a with mean weight Wt-1, where the growth rela- catch-age model. tionship (T6.4.2) produces a mean weight l+kWt-1 one year later. Given the biomass B computed Q=(a, b, g, d, w ¸k, l, q , q ) (T6.6.1) t r 1 2 from (T6.6.3), (T6.6.4) updates the mean weight as R =aS (1 -bgS )1/g (T6.6.2) t t-r t-r the ratio of biomass to number of fish, where the lk+ Wt-1 denominator expresses population numbers as BRtt=+ ()1- d St-1 (T6.6.3) Wt-1 ratios of biomass to mean weight. -1 R S WBÈ t 1 d t-1 ˘ The remaining equations in model 2 describe tt=+-Í ()˙ (T6.6.4) Îwr Wt-1 ˚ the effects of fishing and indexing. As in model 1, Ft = q1Et (T6.6.5) (T6.6.5) assumes that fishing mortality is propor- H 1 e-Ft (T6.6.6) t = - tional to effort. This produces the harvest removal Ct = HtBt (T6.6.7) fraction Ht in (T6.6.6) and catch Ct in (T6.6.7). The St = Bt - Ct (T6.6.8) spawning stock S consists of biomass that re- Ê 1 ˆ t IqBttt=-2 C (T6.6.9) mains after the fishery. Finally, (T6.6.9) relates the Ë 2 ¯ index It to available biomass, assuming that the Surplus Production Models 117

Effort rate Et

T6.6.5

Fishing mortality Ft

Lag 1 Weight Wt T6.6.6

Harvest fraction Ht

Dynamics T6.6.3 and Biomass Bt T.6.6.6T6.6.7 Catch Ct T6.6.4

T6.6.8 T6.6.9

Lag 1 Spawning stock St Index It

Lag r

T6.6.2 Fig. 6.4 Flow diagram for model 2 in Table 6.6, showing the state variables calculated from each Recruitment R equation (T6.6.2)–(T6.6.9). t

index measures conditions halfway through the d=1 - e-M (6.18) fishery. Model 2 assumes a particular annual cycle of associated with natural mortality M. Thus, events: recruitment, fishing, spawning and natural harvest and natural mortality fractions Ht and d mortality. Model variations could examine other have similar relationships (T6.6.6) and (6.18) scenarios, such as concurrent fishing and natural to fishing and natural mortality coefficients Ft mortality, or natural mortality prior to spawning. and M. For small values of mortality, Ht ª Ft and The model includes nine parameters (T6.6.1): pro- dªM. portionality constants (q1, q2) similar to those in Model 2 incorporates features that reflect a model 1, growth parameters (wr, l, k), recruitment diverse historical literature. Schnute (1985) sug- parameters (a, b, g) for (T6.6.2), and a natural mor- gested the recruitment function (T6.6.2), based on tality fraction earlier work by Deriso (1980). As in model 1, the 118 Chapter 6 parameter g selects from a variety of curve shapes, where a limit exists as gÆ0. Special cases g=-1, 0 6.8 EQUILIBRIUM AND and 1 correspond, respectively, to models proposed DYNAMIC RESPONSE by Ricker (1958), Beverton and Holt (1957) and (MODEL 2) Schaefer (1954). Recruitment curves can increase indefinitely (g<-1), ascend toward an asymptote Model 2 has asymptotic properties similar to those

(g=-1), or rise to a maximum value and then de- for model 1. With no fishing (H = 0), the biomass Bt cline (g>-1). Schnute (1985, Figs 2–4) illustrates converges to a carrying capacity B¢. As the harvest the possibilities. fraction H increases from 0, the corresponding We caution the reader not to confuse different equilibrium biomass B(H) declines from B¢. Ana- roles for g in models 1 and 2. For example, the case lytical results for model 2, however, are not as g=1 in model 2 gives a parabolic recruitment simple as those for model 1. Table 6.7 shows curve: R =aS(1 -bS). Similarly, the case g=1 in expressions for equilibrium biomass states associ- model 1 gives a differential equation with a para- ated with H in the control system bolic growth rate: dB/dt =aB[1 - (B/B¢)]. This, how- (W, R, B, C, S. (6.19 ۋ ever, implies a hyperbolic relationship (T6.2.7) H between Bt and Bt+t, analogous to the hyperbolic recruitment curve R =aS/(1 +bS) in model 2 when The three remaining states (E, F, I) in Table 6.6 g=-1. From this point of view, it could be argued have simple relationships to those in the shorter that Schaefer’s combined papers (Schaefer 1954, list (6.19). Equations (T6.7.1)–(T6.7.7) define a 1957) really dealt with a hyperbolic relationship, sequential calculation, in which each equilib- similar to that of Beverton and Holt (1957). At the rium state depends on previous values. Two inter- risk of further confusion, we could add that model mediate values, s and r, have useful biological 1 with g=1 implies a logistic relationship (T6.2.7) interpretations. The quantity s(H) represents a sur- between Bt+t and t. vival fraction from the combined effects of fishing Deriso (1980) and Schnute (1985) proposed and natural mortality, and difference equations similar to those in model 2. Their results follow mathematically from an r()HRHSH= ()() (6.20) age-structured model with three simplifying assumptions. First, natural mortality remains denotes the biomass ratio of recruits to spawners. constant across all years and age classes. Second, the fishery has a uniform, knife-edged selectivity on fish at age r and older. Third, maturity also Table 6.7 Equilibrium values for states in model 2 at a occurs at age r, with fecundity proportional to fixed harvest rate H, where r=R/S. biomass. Originally, these analyses led to second- s(H) = (1 -d) (1 - H) (T6.7.1) order difference equations, with no reference ()1- swr + sl to the mean weight W . Schnute and Richards WH()= (T6.7.2) t 1-ks (1998) extended this work to obtain first-order w 1- s difference equations involving the additional r()H = r (T6.7.3) W 1- H state Wt, which potentially can be measured. g r È Ê r ˆ ˘ Based on the growth model (T6.4.2), their formula- RH()=-Í1 ˙ (T6.7.4) bg Ë a ¯ tion has the simple biological interpretation de- ÎÍ ˚˙ W R scribed above. Mathematically, the linear form of BH()= (T6.7.5) (T6.4.2) allows the growth of individual fish to be wr 1- s C(H) = HB (T6.7.6) accumulated into a similar growth law for the S(H) = B - C (T6.7.7) whole population. Surplus Production Models 119

Table 6.8 Parameters (a, b) for the model 2 expressed in pristine biomass B¢=S¢. As S Æ 0, the recruitment terms of key reference points. function (T6.6.2) predicts that r=R/S Æa. In a viable stock, productivity must be larger at small Given (H*, C*) stock sizes than under pristine conditions, that is, s* = (1 -d) (1 - H*) (T6.8.1) a>r¢. By (6.21), this implies that the parameters ()1- sw**r + sl W* = (T6.8.2) (T6.6.1) for model 2 must satisfy the condition 1-ks* w 1- s* Q* =-Ê1 r ˆ (T6.8.3) dkkdw r ()1-+ Ë W*¯ 1-k s* a > . (6.22) ddlw r +-()1 1 g wr 1- s* È g ()1+ QH**˘ a = 1+ (T6.8.4) W* 1- H* ÎÍ 1- s* ˚˙ Let f=a/r¢ > 1 be the factor that represents the ()1+ QH**2 increase in recruitment productivity from a pris- b = (T6.8.5) ()11- Hs**[]-+g () 1 + QHC *** tine stock S¢ to a small stock S ª 0. Then parameters (B¢, f) can act as surrogates for (a, b), as shown Given (B¢, f) explicitly in Table 6.8. Walters and Bonfil (1999) used a similar approach in their analysis of Pacific ddlwr +-()1 W¢ = (T6.8.6) groundfish. The parameter often has a clearer 11--kd() f biological interpretation than a. Furthermore, wr adf= (T6.8.7) estimates of might be available from recruitment W¢ f records for some species. 1 -g b = ()1- f (T6.8.8) g B¢ The parameter f has no meaning in model 1, where recruitment is not modelled explicitly. However, y=C*/B¢ has meaning in both models. To examine the role of y in model 2, consider a

MSY conditions in model 2 cannot be repre- fixed set of growth parameters (wr, k, l), natural sented analytically in terms of the parameters mortality d, and shape parameter g. Table 6.8 de- (T6.6.1), similar to the results in Table 6.3 for fines the parameter transition from (B¢, f) to (a, b), model 1. However, a reverse relationship is pos- from which Table 6.7 gives the equilibrium catch sible (Schnute and Kronlund 1996; Schnute and curve C(H) in (T6.7.6). This curve defines the Richards 1998). The parameters a and b in (T6.6.1) maximum value C* numerically (although not can be computed analytically from the MSY har- analytically) and thus gives y=C*/B¢. Figure 6.5 vest fraction H* and catch C*. Table 6.8 presents illustrates the results of this calculation for a par- these results in the sequential calculation ticular case of model 2 with various values of g. (T6.8.1)–(T6.8.5). Similarly, in model 1, the sim- Each curve defines a functional relationship from f pler relationship (6.13) expresses a in terms of to y. The inverse relationship allows f to be com- y=C*/ B¢. puted from y. The scale parameter B¢ cancels in the When H = 0, Table 6.7 gives a sequential cal- calculations leading to Fig. 6.5, so that we have culation for the carrying capacity B¢=B(0), based established a numerical equivalence on the parameters (T6.6.1). The same calcula- tion gives pristine values of other model states, y´f´a (6.23) which we denote similarly with a prime accent. In particular, among dimensionless parameters for model 2,

given (d, g, wr, k, l). The first equivalence (6.23) R¢ dkkdw r ()1-+ comes from a curve similar to those in Fig. 6.5, and r¢= = (6.21) S¢ ddlw r +-()1 the second from (T6.8.7). The figure shows that al- though any f>1 is admissible, y has an upper limit corresponds to recruitment productivity of the if g<0. In certain cases, y can theoretically exceed 120 Chapter 6

1 (y=C*/B¢>1), suggesting an annual catch C* greater than the pristine biomass B¢. Such cases 1.000 also require a high f associated with extremely 0.500 large production at small stock sizes. We have already demonstrated that three para- meters (B¢, y, g) govern population dynamics in model 1. Similarly, given the underlying character-

0.100 istics of natural mortality (d) and growth (wr, k, l),

Ψ (6.23) and (T6.8.7)–(T6.8.8) show that the same 0.050 three parameters (B¢, y, g) govern the dynamics of model 2. Figure 6.6 shows catch curves for a case of model 2 with y=0.05. These bear a striking resem- γ 0.010 blance to their counterparts in Fig. 6.3 for model 1. The parameter g plays a similar role in both 0.005 models. One notable difference, however, distin- guishes Figs 6.3(b) and 6.6(b). In model 1, the ratio y=C*/ B¢ is eliminated by scaling the biomass B 1 5 10 50 100 and catch C relative to B¢ and C*, respectively. φ Thus, Fig 6.3(b) does not depend on the choice of y. (Mathematicians can construct a proof from Fig. 6.5 y=C*/B¢ as a function of f=aS¢/R¢ in model 2 (6.12)–(6.13).) The more complex model 2 does not with d=0.1, wr = 0.3kg, w• = 1kg, k=0.7. Curves represent various values of g: -1.5, -1, 0, 1, 3. An arrow share this feature. For example, modal points on indicates the direction of increasing g. Heavy solid and the curves in Fig. 6.6(b) shift closer together as y broken curves indicate special cases g=-1 and g=0, decreases from 0.05 to 0.03. respectively. Our model presentations have stressed the role and biological meaning of key parameters. We con- sider this an important part of model development,

(a) (b) 1.0 1.0

0.8 0.8

0.6 0.6

C/C* 0.4 C/C* 0.4

0.2 γ 0.2 γ

0.0 0.0 0.00.1 0.2 0.3 0.4 0.5 0.6 0.0 0.2 0.4 0.6 0.8 1.0 H B/B'

Fig. 6.6 Relative annual catch C/C* for the equilibrium version of model 2 as a function of (panel (a)) harvest rate

H and (panel (b)) relative biomass B/B¢. The parameters y=0.05, d=0.2, wr = 0.3kg, w• = 1kg, k=0.7 are held fixed in each panel. Curves represent various values of g: -1.5, -1, 0, 1, 3. Arrows indicate the direction of increasing g. Heavy solid and broken curves indicate special cases g=1 and g=0, respectively. Surplus Production Models 121

(a) (b)

100 100

80 80

60 60 Biomass Biomass 40 40

20 20

0 0 0 20 40 60 80 100 0 20406080100 Year Year

Fig. 6.7 Biomass trajectories for models 1 and 2 (panels (a) and (b), respectively) with y=0.06 and g=0, where B¢ is scaled to 100 biomass units. Model 1 uses a standard time unit t=1yr, and model 2 has the additional parameters d=0.03, r = 10yr, wr = 0.3kg, w• = 1kg, k=0.7. Pristine conditions initialize each trajectory, followed by four 25yr periods of constant harvest rates: H = 0.2, 0.07, 0.15, 0.07. Solid curves represent biomass trajectories, and broken rectangular lines indicate asymptotic biomass levels B•(H) associated with the four harvest rates.

because parameters encapsulate information to be tine conditions (Bt = St = B¢ and Ht = 0 for t < 1). estimated from available data (see also Sparre and Model 1 follows an orderly progression of decay or Hart, Chapter 13, this volume). Furthermore, growth towards the relevant equilibrium level three common parameters capture the biological B(H). Model 2 has a similar trajectory, although similarity between models 1 and 2, even though biomass levels Bt go beyond the equilibrium value they have quite different mathematical deriva- B(H) in each 25yr period. This tendency to over- tions. Much of the biological interpretation pre- shoot results from a Ricker recruitment function sented above for model 1 carries over directly to (g=0) with a long recruitment lag (r = 10yr). Also, model 2. In particular, both models have a carrying the stock has a low natural mortality rate (d=0.03) capacity B¢, harvest potential y=C*/B¢, and pro- on fish that grow from wr = 0.3kg to w• = 1kg. ductivity parameter g related to the MSY harvest Thus, both recruitment and growth contribute sig- fraction H* and relative biomass B*/B¢ (Figs 6.3 nificantly to the biomass dynamics in Fig. 6.7(b). and 6.6). Shepherd and Pope (Chapter 8, this volume) also Although the two models have similar equilib- discuss aspects of forecasting. rium properties, model 2 produces more complex Figures 6.7(a) and 6.7(b) illustrate a dynamic fea- dynamic trajectories. Figure 6.7 illustrates the dy- ture common to many fisheries. The decay to- namic response of model 1 (Fig. 6.7a) and model 2 wards a low stock level happens more quickly than (Fig. 6.7b) to a sequence of four harvest fractions H, recovery to a higher level. Particularly with each sustained for a period of 25yr. The figures modern fishing gear, humans can often remove show biomass trajectories Bt in relation to the cor- fish much faster than nature can replace them. responding four equilibrium levels B(H). The para- Response times for decay and recovery obviously meter B¢=100 sets a convenient vertical scale. reflect underlying model parameters, and the abil- Both model trajectories are initialized from pris- ity to estimate these parameters may depend on 122 Chapter 6 contrasting fishing conditions, such as those por- tuted a=4y from (6.13) when g=1. Given the trayed in Fig. 6.7. known data, (6.24) determines the residual et(B¢, y, q1) as a function of the three unknown parameters for t = 1,..., n - 1, and the minimum sum of n-1 2 6.9 ESTIMATION: squares St=1 et can be used to obtain estimates (Bˆ¢ , ˆ ˆ COMPARING MODELS y, q1 ). As discussed by Schnute (1977), (6.24) can be WITH DATA written approximately as

ÊUt+11ˆ q 2y It helps to understand a model well before using logÁ ˜ ª-4y ()EEtt +++1 - ()UU t+ t1 + e t . Ë ¯ qB it to estimate parameters from data. We have Ut 2 1 ¢ devoted most of this chapter to a careful examina- (6.25) tion of models 1 and 2 in the context of surplus pro- duction. Nevertheless, the process of working This implies that parameters could be estimated backwards from data to parameters and states re- by linear regression on quantities known from the mains a very important issue (Hilborn and Mangel data. 1997). In fact, the definition of error in the estima- The formulation (6.24), based on (T6.2.9), as- tion phase can actually dominate the structural sumes that all the error occurs in the process of assumptions in the deterministic model (Schnute incrementing Bt to Bt+t. The effort Et in (6.10) pre- 1985, 1987). A complete explanation of this cisely determines the fishing mortality Ft, and the subject is beyond the scope of this chapter, and CPUE Ut in (6.11) gives an exact index of biomass we confine our discussion to a few illustrative Bt. Thus, the model involves process error in the scenarios. biomass dynamics, but no measurement error in After devising models such as those above, the the data. next question is: what is known and what is An alternative approach would consider mea- unknown? For example, consider Schnute’s inte- surement error in the data but an exact system grated form (T6.2.9) of model 1 with g=1. Usually, dynamics. Consider the general formulation of none of the quantities in this equation are known, model 1 (g unrestricted) in the context of small except the time interval t, say, t=1yr. As discussed time steps t. For example, suppose that t is mea- earlier, suppose that annual effort and CPUE data sured in years, but t=1 week = 1/52yr. Let the data

(Et,Ut) are observed for a series of years t = 1,..., n. consist of weekly values (Et, Ct) with indices t in Equations (6.10)–(6.11) allow the unknown quanti- fractional time, plus an annual fishery indepen- ties (Ft, Bt) to be replaced with known data, along dent index It with indices confined to integers t. with the fixed parameter q1. This suggests the sto- The estimation process starts with assumed trial chastic model values for the parameters (T6.1.1) and an initial

biomass B1 at the start of the first week. Sequential deterministic calculations from (T6.1.2) and Ê 4yUt+1 ˆ exp 1 (T6.1.8) produce a complete trajectory of states F Á ˜ - ()4y-qE11t+ t Ë qB1 ¢ ¯ 4y - qE1 t e -1 = and Bt for each week t. Introducing measurement ()4y-qE1 t Ê 4yUt ˆ 4y - qE11t+ e -1 expÁ ˜ -1 error into (T6.1.4) and (T6.1.5) gives Ë qB1 ¢ ¯ 4 U e ÈÊ y t ˆ˘ et C = F B e 1t, (6.26) ¥--expÍÁ4y qE1 t ˜˙e , (6.24) t t t ÎË qB1 ¢ ¯˚ e2t It = q2 Bte , (6.27) where et is a normal variate with mean 0 and vari- 2 ance s . In deriving (6.24) from (T6.2.9), we have where normal residuals eit have mean 0 and vari- 2 adopted a time-scale in years with t=1 and substi- ances si (i = 1,2). Parameters and the initial state B1 Surplus Production Models 123

could then be estimated by minimizing the sum of vations It and At. They also include process error in squares the recruitment equation, which is somewhat dif- ferent from (T6.6.2) here, and they add a vector of recruitment values R to the list of model parame- n t s2 n t 2 1 2 , ters. Because their analysis involves both process ÂÂe1,mt + 2 e2t (6.28) m==1 s2 t 1 and measurement error, additional assumptions are required. They choose to fix the ratio of process based on data from n years of the fishery. The vari- error to the total (measurement plus process) error. 2 2 ance ratio s1/s2 could be set arbitrarily to place It is well known for these models that the greater emphasis on the catch or index data. choice of error structure can significantly influ- One significant weakness to the scenarios ence the outcome (Ludwig and Walters 1981; above lies in the assumption that fishing effort Schnute 1991; Pella 1993; Polachek et al. 1993).

Et precisely determines the fishing mortality Ft. For example, with a given deterministic model and Typically, effort is poorly known and inconsistent- data set, different error structures can produce ly related to fishing mortality. Catch data tend to mutually exclusive confidence intervals for the be better defined, and we prefer models that re- same parameter or reference point (Schnute 1989). quire catch to be removed directly from the bio- Intuitively, a model filters a signal from noisy data. mass as an overt control, rather than as an Deterministic assumptions, such as those in mod- observation related to the hypothetical effort con- els 1 or 2, specify the desired signal. Error assump- trol. Although model 1 can be reformulated from tions, similar to (6.24), (6.26) and (6.27), define the this point of view, the difference equations in noise. With highly variable data, the perceived sig- model 2 make this approach more straightforward. nal obtained by parameter estimation can depend Consider a scenario in which the data consist as much on error assumptions as on the structural only of the annual catch Ct and an abundance model itself (Schnute 1987). index It for years t = 1,..., n. For simplicity, as- sume that all states are initialized at pristine levels (R¢, S¢=B¢, W¢), as in the calculations for Fig. 6.7(b). 6.10 MODEL EXTENSIONS

Then, given the parameters (B¢, y, g, d, wr, k, l, q2) and the catch history Ct, Table 6.6 gives equations Some of the limitations of models 1 and 2 can be for a deterministic updating of all states Rt, Bt, Wt, addressed formally through model extensions. St and It. Furthermore, Ht = Ct/Bt is also deter- The most common approach is to add new states or mined from (T6.6.7). Equations (T6.6.5)–(T6.6.6) allow parameters to become time-varying states. do not enter this calculation. Similar to (6.27), the For example, a variable carrying capacity B¢ might parameters for model 2 could be estimated by com- be represented as a state Bt¢. The model then re- paring observations It with the model predictions quires an additional assumption to describe the from (T6.6.9). Usually some parameters, such as time trajectory of Bt¢, such as the random walk those related to growth (wr, k, l), would be fixed at ht values known from biological samples and other Bt¢+1 = Bt¢e (6.29) external data sources. A variant of this analysis might use observed mean weights Wt, so that com- where normal residuals ht have mean 0 and vari- 2 parisons of observed and predicted values of this ance s3. Alternatively, (6.29) could relate Bt¢ to an state would also enter the estimation procedure. external environmental variable or an abundance Richards and Schnute (1998) detail the applica- index for a competitor species. The addition of tion of maximum likelihood theory for a model state variables does, however, increase model similar to model 2, except that they replace complexity and parameter uncertainty. (T6.6.4) with an update equation for the mean age For illustration, consider model 2 with n years

At. They assume measurement error in the obser- of data (Ct, It). If the growth parameters are as- 124 Chapter 6 sumed known and the model is initialized with Schnute et al. (2000) use Bayes methods to exam- pristine conditions, then the 2n observed data val- ine management options for a fishery modelled ues would be used to estimate the five parameters with the stock–recruitment relationship (T6.6.2).

(B¢, y, g, d, q2). To allow time variation in the carry- Freeman and Kirkwood (1995) and Reed and ing capacity B¢, the model requires an additional n Simons (1996) investigate Kalman filters as more unobserved states Bt¢that replace the original para- classical estimation tools in various production meter B¢. In this case, the 2n data values must be models intended for use with CPUE data. used to estimate n + 4 unknowns. The sum of Statistical theory cannot circumvent funda- squares function (6.28) would contain an extra mental limitations in the observed data. Fisheries 2 2 n-1 2 term (s1/s3) St=1 ht associated with (6.29). As s3 Æ data sets are notoriously noisy, leading to ambigu- 0, this term imposes a high penalty that causes ous inferences. Meaningful conclusions about a ˆ residuals ht to become small and estimates Bt¢ to dynamic system require observations from periods converge to the estimate Bˆ¢ obtained under the of substantial change, such as the biomass shifts assumption of constant carrying capacity. illustrated in Fig. 6.7. Fishery data often lack this An extension to multi-species models would contrast. Many fisheries begin to operate on a pris- similarly require the use of additional states, tine biomass and progress along a one-way trip the biomass Bit for each species i. Model 1 has downward towards economic collapse (Hilborn a classical extension in this direction, where and Walters 1992, pp. 312–19). Do the data allow the Lotka–Volterra differential equations extend managers to recognize the problem soon enough to model 1 with g=1 to a predator–prey relationship avert disaster? According to Walters (1998, p. 284), between two species (Pauly and Christensen, ‘the sad fact is that most stock assessment systems Chapter 10, this volume). Quinn and Deriso (1999, are just not up to the task of providing accurate p. 9) discuss the generalization to an arbitrary enough stock size estimates to be safe in the face of number of species. Similar extensions could, of fishing technologies that can get the quota even if course, be incorporated into model 2. For example, stock size is much lower than estimated’. consider two independent versions of this model, one for the predator and the other for the prey species. Additional equations could link these 6.11 CONCLUSIONS models by relating the natural mortality, growth and recruitment of each species to the abundance Despite the known limitations of fishery models, of the other. Unfortunately, as discussed by we have developed them here in mathematical de- Hilborn and Walters (1992, pp. 437–8), realistic tail. We have also described a design philosophy data generally lack the contrast needed to estimate that allows models to be tailored to specific cir- all the necessary parameters. cumstances. Fishery models begin with an arith- For simplicity, we have suggested a sum of metic that tries to capture the essential concept of squares objective function to estimate parameters surplus production. Schnute and Richards (2001) in the examples above. More generally, likelihood coin the word ‘fishmetic’ (rhymes with arithmetic) (Edwards 1972; Schnute 1994) or Bayes posterior to represent uncertainty in the conversion of arith- (Gelman et al. 1995; Punt and Hilborn 1997) analy- metic to practical fishery models. That paper, writ- ses give a complete description of parameter esti- ten concurrently with this chapter, identifies key mates and their uncertainty. The process of reasons why models can fail. Readers may wonder converting parameters to states fits particularly how the same authors can present the elaborate well into the Bayes framework of conditional dis- mathematics of this chapter and yet, in another tributions and hierarchical models (Gelman et al. forum, identify severe problems with its applica- 1995, chap. 5). Two recent papers (Meyer and tion. We believe that the surplus production para- Millar 1999a,b) illustrate the Bayes approach in digm, portrayed by example in Fig. 6.1, lies at the contexts similar to models 2 and 1, respectively. heart of any rational theory of fishing. Without Surplus Production Models 125 surplus production, no fishery could be sustained. and the Kalman filter for predicting consequences of Anyone interested in pursuing criticisms or varia- fishery actions. In: Proceedings of the International tions of this paradigm should understand it fully. Symposium on Management Strategies for Exploited Fish Populations, Alaska Sea Grant College Program The sceptical modeller uses production models as Report, no. 93–02. University of Alaska, Fairbanks, pp. tools for understanding and communicating the 571–93. inevitable limits to fishery harvest. Pella, J.J. and Tomlinson, P.K. (1969) A generalized stock production model. Inter-American Tropical Tuna Commission Bulletin 13, no. 3. Polachek, T., Hilborn, R., and Punt, A.E. (1993) Fitting REFERENCES surplus production models: comparing methods and measuring uncertainty. Canadian Journal of Fisheries Beverton, R.J.H. and Holt, S.J. (1957) On the Dynamics of and Aquatic Sciences 50, 2597–607. Exploited Fish Populations. London: Ministry Agri- Punt, A.E. and Hilborn, R. (1997) Fisheries stock assess- culture, Fisheries and Food, Fisheries Investigation ment and decision analysis: the Bayesian approach. Re- Series 2, no. 19. views in Fish Biology and Fisheries 7, 35–63. Deriso, R.B. (1980) Harvesting strategies and parameter Quinn, T.J., II and Deriso, R.B. (1999) Quantitative Fish estimation for an age-structured model. Canadian Dynamics. New York: Oxford University Press. Journal of Fisheries and Aquatic Sciences 37, 268–82. Reed, W.J. and Simons, C.M. (1996) Analyzing catch- Edwards, A.W.F. (1972) Likelihood: An Account of the effort data by means of the Kalman filter. Canadian Statistical Concept of Likelihood and its Application Journal of Fisheries and Aquatic Sciences 53, 2157–66. to Scientific Inference. Cambridge: Cambridge Uni- Richards, L.J. and Schnute, J.T. (1992) Statistical models versity Press. for estimating CPUE from catch and effort data. Cana- Fox, W.W. Jr. (1970) An exponential surplus-yield model dian Journal of Fisheries and Aquatic Sciences 49, for optimizing exploited fish populations. Transac- 1315–27. tions of the American Fisheries Society 99, 80–8. Richards, L.J. and Schnute, J.T. (1998) Model complexity Fox, W.W. Jr. (1975) Fitting the generalized stock produc- and catch-age analysis. Canadian Journal of Fisheries tion model by least-squares and equilibrium approxi- and Aquatic Sciences 55, 949–57. mation. US Fishery Bulletin 73, 23–37. Ricker, W.E. (1958) Handbook of computations for bio- Freeman, S.N. and Kirkwood, G.P. (1995) On a structural logical statistics of fish populations. Bulletin of the time series method for estimating stock biomass and Fisheries Research Board of Canada, no. 119. recruitment from catch and effort data. Fisheries Re- Schaefer, M.B. (1954) Some aspects of the dynamics of search 22, 77–98. populations important to the management of the com- Gelman, A., Carlin, J.B., Stern, H.S. and Rubin, D.B. mercial marine fisheries. Inter-American Tropical (1995) Bayesian Data Analysis. London: Chapman & Tuna Commission Bulletin 1, 26–56. Hall. Schaefer, M.B. (1957) A study of the dynamics of the fish- Hilborn, R. and Mangel, M. (1997) The Ecological Detec- ery for yellowfin tuna in the eastern tropical Pacific tive: Confronting Models with Data. Princeton: Ocean. Inter-American Tropical Tuna Commission Princeton University Press. Bulletin 1, 245–85. Hilborn, R. and Walters, C.J. (1992) Quantitative Fish- Schnute, J.T. (1977) Improved estimates for the Schaefer eries Stock Assessment: Choice, Dynamics and Un- production model: theoretical considerations. Journal certainty. New York: Chapman & Hall. of the Fisheries Research Board of Canada 34, Ludwig, D. and Walters, C.J. (1981) Measurement errors 583–603. and uncertainty in parameter estimates for stock Schnute, J.T. (1981) A versatile growth model with statis- and recruitment. Canadian Journal of Fisheries and tically stable parameters. Canadian Journal of Fish- Aquatic Sciences 38, 711–20. eries and Aquatic Sciences 38, 1128–40. Meyer, R. and Millar, R.B. (1999a) Bayesian stock assess- Schnute, J.T. (1985) A general theory for the analysis of ment using a state–space implementation of the delay catch and effort data. Canadian Journal of Fisheries difference model. Canadian Journal of Fisheries and and Aquatic Sciences 42, 414–29. Aquatic Sciences 56, 37–52. Schnute, J.T. (1987) Data uncertainty, model ambiguity, Meyer, R. and Millar, R.B. (1999b) BUGS in Bayesian and model identification. Natural Resource Modeling stock assessments. Canadian Journal of Fisheries and 2, 159–212. Aquatic Sciences 56, 1078–86. Schnute, J.T. (1989) The influence of statistical error Pella, J.J. (1993) Utility of structural time series models in stock assessments: illustrations from Schaefer’s 126 Chapter 6

model. Canadian Special Publication of Fisheries and Schnute, J.T. and Richards, L.J. (1998) Analytical Aquatic Sciences 108, 101–9. models for fishery reference points. Canadian Journal Schnute, J.T. (1991) The importance of noise in fish of Fisheries and Aquatic Sciences 55, 515–28. population models. Fisheries Research 11, 197–223. Schnute, J.T. and Richards, L.J. (2001) Use and abuse of Schnute, J.T. (1994) A general framework for developing fishery models. Canadian Journal of Fisheries and sequential fisheries models. Canadian Journal of Aquatic Sciences 58, 10–17. Fisheries and Aquatic Sciences 51, 1676–88. Walters, C.J. (1998) Designing fisheries management sys- Schnute, J.T., Cass, A. and Richards, L.J. (2000) A tems that do not depend upon accurate stock assess- Bayesian decision analysis to set escapement goals for ment. In: T.J. Pitcher, P.J.B. Hart and D. Pauly (eds) Fraser River sockeye salmon (Oncorhynchus nerka). Reinventing Fisheries Management. London: Kluwer Canadian Journal of Fisheries and Aquatic Sciences Academic Publishers, pp. 279–88. 57, 962–79. Walters, C.J. and Bonfil, R. (1999) Multispecies spatial as- Schnute, J.T. and Kronlund, A.R. (1996) A management sessment models for the British Columbia groundfish oriented approach to stock recruitment analysis. trawl fishery. Canadian Journal of Fisheries and Canadian Journal of Fisheries and Aquatic Sciences Aquatic Sciences 56, 601–28. 53, 1281–93. 7 Dynamic Pool Models I: Interpreting the Past using Virtual Population Analysis

J.G. SHEPHERD AND J.G. POPE

7.1 INTRODUCTION: THE years, and denoted by the index a, as a function of DYNAMIC POOL CONCEPT time, also usually expressed in years, and denoted by the index y. Dynamic pool methods seek to esti- The concept of the dynamic pool as a model for fish mate P(y,a) from available data, and predict its populations is very simple. The population con- future evolution under various assumptions about sists of all individuals alive at any time. This popu- natural and anthropogenic effects, especially lev- lation is continuously reduced in size by deaths, els of fishing. The data to be analysed are generally due to natural mortality or to fishing, and is aug- some estimates of catch-at-age C(y,a), possibly dis- mented by recruitment of young fish. This consti- aggregated between fleets and indexed by f (Sparre tutes the dynamic pool. The concept was implicit and Hart, Chapter 13, this volume), one or more in the earliest attempts to model their dynamics indices of abundance u(y,a,f ) such as CPUE (catch quantitatively, but it was developed to fruition in per unit effort) estimates for a number of fleets the work of Beverton and Holt (1957). It is general- and/or research vessel surveys. Estimates of the ly regarded as distinct from the picture underlying rates of mortality due to natural causes, M(a), stock-production models (Schnute and Richards, which is often expressed as a function of age and Chapter 6, this volume), although these too in- sometimes also of time, and that due to fishing, volve a dynamic balance between production (due F(y,a), as well as their sum, the total mortality rate to recruitment and individual growth) and deple- Z(y,a), are the natural variables used to describe tion due to mortality, because more attention is the fate of the fish. paid to the fate of individuals within the ‘pool’, and the consequent structure and dynamics of the population. 7.2 THE DISTINCTION This description naturally focuses attention on the processes of reproduction with its consequent BETWEEN RETROSPECTIVE recruitment of young fish to the population, indi- ANALYSIS AND vidual growth and maturation, and death. These FORECASTING are usually described as a function of the age of the individuals involved, and there is therefore a close Within the general class of dynamic pool methods, correspondence between the dynamic pool con- there tends to be a fairly clear separation between cept and age-structured methods of analysis and the techniques for the analysis of past data, to esti- prediction. The natural description of the popula- mate the current state and structure of the popula- tion is therefore in terms of the population number tion, and those for forecasting its future evolution. P(y,a) of fish in each age class, usually measured in The former are exemplified by the ubiquitous 128 Chapter 7 technique known as Virtual Population Analysis especially when the retrospective analysis and the (VPA), which is the subject of this chapter. The forecast are themselves integrated within a formal latter include both short-term catch forecast management procedure. These too have undoubt- methods, and long-term analyses such as that of ed advantages, but when they are applied it can be- yield-per-recruit (introduced by Beverton and Holt come very difficult to figure out what is actually 1957), which are the subject of Chapter 8. There is going on, when surprising or controversial results no essential reason for this separation, and meth- are obtained. ods certainly exist in which both analyses are con- ducted together within a unified framework (for an up-to-date account of such methods see Quinn and 7.3 VIRTUAL POPULATION Deriso 1999). There are advantages and disad- ANALYSIS (VPA): vantages in both approaches. However, it is a re- THE BASICS grettable but undeniable fact that fish stock assessments sometimes go wrong, or at least 7.3.1 The virtual population become the subject of intense debate, and it is and cohort analysis not unusual for assessment scientists to spend The original concept of the virtual population – a considerable time trying to figure out what hap- minimum estimate of the size of a year class at pened, and why, either in reality, or in their analy- any age, obtained by summing all the catches sis, or both. It is therefore usually helpful to have as subsequent to that age – was introduced by Fry clear a picture as possible of the actual (estimated) (1949). The idea did not become popular, however, current state of the stock, and the information on until the 1970s, following the demonstration by which this estimate is based, as well as of its ex- Gulland (1965) that one could also allow for natur- pected future evolution, and the principal factors al mortality to obtain a more realistic estimate. determining this in both the short term and in Gulland’s working paper was not generally avail- the long term. Moreover, it is often necessary to able for many years, but is now included in the include in the analysis some assumptions about compilation by Cushing (1983). changes, either those to be imposed as a result of Gulland’s method depended on the iterative so- management action such as reductions of fishing lution of the conventional catch equation for each mortality, or changes of the exploitation pattern age and year (equation 13.19 in Sparre and Hart, given by the set of values of F on all ages in any year, Chapter 13, this volume), i.e. or to be expected from natural causes such as changes of climate regimes. It is helpful if such assumptions are exposed as clearly as possible for CFZ= ()[]1-- exp() ZP (7.1) critical evaluation and debate. Both of these pur- poses, as well as the explanation of the workings of the analyses, are assisted to some extent by treat- where we have omitted the indexes y and a for clar- ing the retrospective analysis and the forecasting ity. This was a laborious and expensive calculation process separately. We have therefore adopted this when done on an electromechanical calculator, approach, which also reflects the structure of and it was not trivial even on an early digital com- much of the software available for these analyses, puter. The so-called ‘cohort analysis’ approxima- in the descriptions of these methods given here and tion introduced by Pope (1972) – also reprinted in in Chapter 8. However, there are also advantages in Cushing (1983) – was therefore a significant ad- using a unified approach to analysis and forecast- vance. It is in fact a remarkably good approxima- ing, and our approach is not intended to detract tion – good enough for all practical purposes. Now from these, except in so far as critical features of that computation is much cheaper, however, its the analysis may sometimes be more difficult to main virtue is that it exposes very clearly indeed discern in such unified procedures. This applies the essence of virtual population analysis (VPA) Dynamic Pool Models I: Interpreting the Past 129

80000 The fishing mortality values can of course be calculated immediately once the populations have 70000 been estimated, since Natural deaths 60000 Adjusted catch Zya(),,,= 1n[] Pya() Py()++ 1 a 1 (7.3) Survivors 50000 and

40000 Fya(),,= Zya()- Ma() . (7.4)

30000 Clearly also, in order to estimate the total popu- lations, we need to have and to use the total inter- Number of fish in the cohort 20000 national catch numbers for C(y,a) – partial data from a few fleets is not enough, although the pos- 10000 sibility of relaxing this restriction is discussed at the end of this chapter. Finally, it is clear that one 0 123456 needs some estimate for terminal population size Age at the end of the final year for which a catch esti- mate is available, in order to start the calculation, Fig. 7.1 Accumulation of the virtual population. which is carried out retrospectively, as implied by equation (7.2). ‘Exact’ VPA, based directly on the solution of equation (7.1) (with equations (7.3) and (7.4)) can of as a technique, and for that reason we employ it course be carried out instead of the cohort approxi- throughout this chapter. mation, and this is in fact the usual operational Pope’s formula practice. A large number of algorithms for the iterative solution of this equation have been pro- Pya(),= exp[] Ma()2 Cya() , posed, including the standard Newton–Raphson + exp[]Ma() Py()++11 , a (7.2) method (see for example Stephenson 1973). Given the terminal population, all the other can be derived, using a little ingenious algebra, as population-at-age estimates can thus be computed an approximation to (7.1). It is, however, also the immediately. And given these, all the fishing obvious result of assuming that all the catch is mortality values can then also be calculated. There taken instantaneously half-way through the year. are no fitting procedures involved, no residuals, no One is simply raising the number of survivors to lack of fit, and no degrees of freedom. VPA is allow for natural mortality over a full year, and the therefore in no real sense a statistical modelling number of fish caught to allow for natural mortal- procedure (G. Gudmundsson, personal communi- ity over only half the year. Clearly, multipliers cation). The catch data are accepted as though they other than exp (M/2) could also be applied if the were exact, and entered directly into the calcula- catches were concentrated at some other time of tion. VPA is therefore best thought of as a transfor- year. In any case, this formulation shows that VPA mation of the data. The catches are converted into is no more and no less than a method for estimating a set of equivalent population estimates, which the population size of a cohort at any age, by ac- may in turn be converted into the equivalent fish- cumulating the subsequent catches, with adjust- ing mortalities. Any errors in the original catch-at- ment for natural mortality. The process, which is age data feed through directly into the population normally applied sequentially for all age groups be- and fishing mortality estimates. longing to the same cohort, is illustrated in Fig. 7.1. Thus understood, VPA is a very quick and con- 130 Chapter 7 venient procedure for turning indigestible catch- is usually referred to as convergence of the VPA, at-age data into things that are more easily under- and is discussed in detail by Pope (1972). As a rule stood, at least by those who have been suitably of thumb, estimates may be regarded as more or trained, namely fishing mortalities and estimates less converged if the cumulative fishing mortality, of year-class strength. Its virtues and vices will be from the oldest age in the cohort back to that in discussed in more detail later on. Here it is neces- question, is greater than one, but not otherwise. sary to stress one vital point: the transformation The converse is also true, that running a VPA performed by VPA is not unique. For every choice calculation forwards in time leads to divergent of terminal population, and there are infinitely estimates for the older ages and years. In practice many such choices, there is a resultant and differ- this means that estimates of terminal F or popula- ent transformation: a few of the possibilities are tion obtained in this way are highly variable, and illustrated in Fig. 7.2. In addition, with no residuals very sensitive to the initial estimates of year-class and no lack of fit, there is no way to choose among strength used to start the calculation. They may them. All are equally good (exact) fits to the data. in fact easily become infeasible (zero or negative) This is the central problem of VPA, that there are for what are a first sight quite plausible initial as- infinitely many possible ‘solutions’ for each and sumptions. Given sufficient precision in the calcu- every cohort. To select among these the user must lation, forwards and backward VPA are of course supply an estimate of the terminal population, or, exactly equivalent calculations, but in practice, equivalently, the terminal fishing mortality, and because of the sensitivity to starting assumptions, there is no objective way of doing this based on the the retrospective calculation is robust, whereas catch-at-age data alone. the prospective one is not, and it is useful only as a Additional information is therefore required. technical device in some special circumstances. This may take the form of additional assumptions, or additional data. In the former case one is in ef- 7.3.2 Exploratory analysis using fect specifying a stronger and less highly parame- trial VPAs terized model. This is the approach adopted by separable VPA, discussed in Section 7.4. It is, The retrospective ‘convergence’ property of VPA however, only partially successful, since serious means that, even without additional information, indeterminacy remains. The second approach, ad- it may often be used to obtain some idea about fish- ducing more data, is typified by the use of CPUE ing mortality a few years ago, and the size of all but and survey abundance data for ‘tuning’ the VPA, the most recent year classes, provided fishing and is discussed in Sections 7.5 and 7.6. mortality is not too small. Indeed, there are few Before leaving the subject of ‘conventional’ (un- professional analysts who, given a suitable set of tuned) VPA, it is important to recognize some of its catch-at-age data, would not carry out some form essential features. A first important point is that of trial VPA just to get some rough idea of what is, natural mortality, M, is taken to be known, and is or rather was, going on. Such a trial analysis must not and cannot be estimated by the analysis. M is usually be based on rather arbitrary assumptions usually taken to be given as a constant or as a func- concerning terminal F. We recommend setting tion of age, but the analysis is easily generalized trial terminal Fs to a high value (of the order of 1) should values of M as a function of both year and in such an analysis, to ensure rapid convergence. age be available. VPA is also essentially a retro- This can conveniently be achieved, especially spective technique, not only because that is the when using cohort analysis, by setting the termi- most convenient way in which to carry out the nal population equal to the terminal catch num- calculations, but because estimates of fishing mor- ber, or some fraction thereof. The value of terminal tality and population size for earlier years and F so created depends on the level of natural mortal- younger ages become progressively less sensitive ity, but is satisfactorily high (0.644 when M is to the starting assumptions about terminal F. This taken as 0.2). Examples of such a trial analysis, for a Dynamic Pool Models I: Interpreting the Past 131

Table 7.1 North Sea cod: catch number at age, 1990–5.

Age Year

1990 1991 1992 1993 1994 1995

1 11841 13628 27967 4814 16173 16465 2 54692 23571 32216 55560 25195 63654 3 11994 16840 8697 11409 21118 12932 4 4360 3319 4995 3211 3078 5296 5 2462 1393 1057 1578 862 800 6 304 1032 479 430 513 283

Table 7.2 North Sea cod: population estimates from trial cohort analysis with high terminal F (0.644).

Age Year

1990 1991 1992 1993 1994 1995 1996

1 69619 87699 163175 84215 198758 38307 0 2 97245 46285 59471 108291 64593 148096 16465 3 21301 30130 16567 19540 38389 30087 63654 4 8361 6587 9431 5695 5675 12322 12932 5 5654 2901 2390 3202 1757 1861 5296 6 707 2401 1114 1000 1194 658 800 7 304 1032 479 430 513 283

Table 7.3 North Sea cod: fishing mortality estimates setting the terminal populations equal to double from trial cohort analysis with high terminal F (0.644). the final catch number, are given in Tables 7.4 and 7.5. Age Year The time trends of average fishing mortality 1990 1991 1992 1993 1994 1995 and year-class strength from these analyses are plotted in Figs 7.2 and 7.3, which illustrate the con- 1 0.208 0.188 0.210 0.065 0.094 0.644 siderable difference between these equally feasible 2 0.972 0.827 0.913 0.837 0.564 0.644 interpretations of the same data, particularly (but 3 0.974 0.962 0.868 1.036 0.936 0.644 not only) for recent years. 4 0.859 0.814 0.880 0.976 0.915 0.644 5 0.656 0.757 0.671 0.787 0.781 0.644 6 0.644 0.644 0.644 0.644 0.644 0.644 7.3.3 The Judicious Averaging Method (JAM) If (and only if) one is prepared to assume that things set of ICES data for the North Sea cod (Gadus such as fishing mortality and the exploitation pat- Morhua) stock from 1990 to 1995 (Anon. 1991) are tern have not changed very much recently, a trial given in Tables 7.1, 7.2 and 7.3, using the fairly analysis can also give some idea of what is going on high estimate of terminal F obtained in this way. now, as well as in the past, but the validity of the The results of an alternative trial analysis, interpretation depends entirely on the assumption using a lower estimate of terminal F, obtained by of constancy. One is really relying on additional 132 Chapter 7

Table 7.4 North Sea cod: population estimates from trial cohort analysis with low terminal F (0.373).

Age Year

1990 1991 1992 1993 1994 1995 1996

1 70558 89874 174962 107779 293719 58417 0 2 98639 47054 61251 117941 83886 225843 32930 3 22258 31272 17197 20998 46289 45882 127308 4 9234 7371 10366 6210 6869 18790 25864 5 7193 3615 3032 3967 2179 2838 10592 6 1079 3662 1699 1526 1820 1004 1600 7 608 2064 958 860 1026 566

Table 7.5 North Sea cod: fishing mortality estimates tively rerun the VPA, inserting terminal Fs based from trial cohort analysis with low terminal F (0.373). on the average over several (usually about five) pre- vious years, for each age in the last year, and the Age Year average over (say) the five next younger age groups 1990 1991 1992 1993 1994 1995 for the oldest age group in each year. This formalizes the assumptions being made, 1 0.205 0.183 0.194 0.051 0.063 0.373 i.e. approximate constancy of F in recent years, and 2 0.949 0.807 0.871 0.735 0.403 0.373 approximate flatness of the exploitation pattern 3 0.905 0.904 0.819 0.917 0.702 0.373 on the oldest ages, and creates a workable method 4 0.738 0.688 0.760 0.847 0.684 0.373 of analysis sometimes known as the JAM method. 5 0.475 0.555 0.487 0.579 0.575 0.373 The acronym may be expanded either informa- 6 0.373 0.373 0.373 0.373 0.373 0.373 tively as the Judicious Averaging Method, or cyni- cally as Just Another Method, according to taste. An example of such an analysis on the same North information from a stronger model, based on the Sea cod data set is given in Tables 7.6 and 7.7, and assumption of more or less constant fishing mor- the resulting fishing mortality and year-class tality, as described above. This procedure can of strength estimates are included in Figs 7.4 and course be automated quite easily. One can itera- 7.5. This third interpretation of the data is less

0.9 0.8 0.7 0.6 0.5

(ages 2 to 6) 0.4 F 0.3 Mean 0.2 0.1 Fig. 7.2 Fishing mortality 0 estimates from two trial VPA 1989 19901991 1992 1993 1994 1995 1996 analyses. ᭿ represents higher F; Year ᭜ represents lower F. Dynamic Pool Models I: Interpreting the Past 133

350000

300000

250000

200000

150000

100000 Population number at age 1 50000 Fig. 7.3 Year-class strengths from two trial VPA analyses. 0 ᭝ represents low F; ᭿ represents 1989 1990 1991 1992 1993 1994 1995 high F. Year

Table 7.6 North Sea cod: population estimates from JAM method.

Age Year

1990 1991 1992 1993 1994 1995 1996

1 69619 87329 158575 73055 165790 122050 2 97245 46285 59167 104524 55456 121103 85028 3 21301 30130 16567 19292 35305 22607 41554 4 8361 6587 9431 5695 5472 9797 6807 5 5654 2901 2390 3202 1757 1695 3229 6 707 2401 1114 1000 1194 658 664 7 304 1032 479 430 513 283

Table 7.7 North Sea cod: fishing mortality estimates not thereby guaranteed to be closer to the truth. from JAM method. This will only happen if the assumptions made happen to be correct. The validity of these assump- Age Year tions is not known and cannot be tested on these 1990 1991 1992 1993 1994 1995 data alone. In the absence of any other relevant informa- 1 0.208 0.189 0.217 0.076 0.114 0.161 tion, however, these assumptions may be reason- 2 0.972 0.827 0.921 0.885 0.697 0.861 able null hypotheses, to be accepted until 3 0.974 0.962 0.868 1.060 1.082 0.989 disproved. They are in fact weaker than the as- 4 0.859 0.814 0.880 0.976 0.972 0.900 sumptions made in some alternative analyses of 5 0.656 0.757 0.671 0.787 0.781 0.730 the same data. If one were to average over years, 6 0.644 0.644 0.644 0.644 0.644 0.644 and plot and analyse the resulting catch curve (see Gulland 1983, Section 4.3.3), then one would be arbitrary than the two trial analyses, since it is implicitly assuming that fishing mortality is con- based on clearly stated assumptions about the be- stant over the whole period in question, that the haviour of fishing mortality. It depends on a rather exploitation pattern is flat for all fully recruited more restrictive model for fishing mortality, but is ages, and that recruitment has been constant, or at 134 Chapter 7

1.2

1.0

0.8

0.6 1 2 3 0.4 4

Fishing mortality (ages 1 to 4) 0.2

0 1989 1990 1991 1992 1993 1994 1995 Fig. 7.4 Fishing mortality Year estimates from the JAM method.

180000 1 160000 2 3 140000

120000

100000

80000

60000

40000 Population numbers at ages 1 to 3 20000 0 1989 1990 1991 1992 1993 1994 1995 1996 Fig. 7.5 Population numbers from Year the JAM method.

least trend-free for the whole period. These are vides no evidence about them. As will be discussed much stronger assumptions, which most analysts below, it is also a method which is very sensitive to would, however, probably make quite cheerfully errors in the catch data, which may seriously affect in sufficiently desperate circumstances. the population estimates obtained. A similar inter- The strong assumptions implied by the JAM pretation of the data, based on similar assump- method need not therefore forbid its use as an ex- tions, which is less sensitive to these errors can in ploratory tool. However, one should remember fact be constructed using separable VPA, and this that it is based on assumptions concerning the is preferable, particularly if the analysis is to form things one would really like to determine, particu- the basis of a catch forecast. Finally, it should be larly the recent trend of fishing mortality, and pro- noted that it is not appropriate to carry out a catch Dynamic Pool Models I: Interpreting the Past 135

Table 7.8 North Sea cod: population estimates from cohort analysis with implausible terminal F.

Age Year

1990 1991 1992 1993 1994 1995 1996

1 72438 86612 198535 72433 483640 28252 2 96548 48593 58581 137241 54947 381337 8233 3 24172 29559 18456 18812 62091 22190 254616 4 7925 8938 8964 7242 5078 31727 6466 5 10272 2543 4315 2819 3023 1373 21184 6 522 6182 822 2576 880 1695 400 7 152 4128 240 1720 257 1132

curve analysis to determine starting assumptions Table 7.9 North Sea cod: fishing mortality estimates for a VPA, whether using the JAM method or other- from cohort analysis with implausible terminal F. wise, because the assumptions made are substan- Age Year tially stronger than those needed for the VPA itself, and are not more likely to be true. 1990 1991 1992 1993 1994 1995

7.3.4 Trial VPA – its virtues and 1 0.199 0.191 0.169 0.076 0.038 1.033 2 0.984 0.768 0.936 0.593 0.707 0.204 its vices 3 0.795 0.993 0.736 1.109 0.471 1.033 A trial VPA is a very simple procedure enabling a 4 0.937 0.528 0.957 0.673 1.108 0.204 5 0.308 0.930 0.316 0.964 0.378 1.033 quick preliminary analysis of a set of catch-at-age 6 1.033 0.204 1.033 0.204 1.033 0.204 data to be carried out. Even the Judicious Averag- ing Method for setting the terminal F values relies on weaker assumptions than alternative methods such as catch curve analysis, and the whole thing can easily be set up using a spreadsheet program 7.4 SEPARABLE VPA on a microcomputer, particularly if one makes use 7.4.1 Separability of fishing of Pope’s cohort analysis algorithm (equation 7.2). mortality The calculation can be done with only (say) 10 years and 10 ages – only 100 numbers – and survey The ‘solutions’ of ordinary VPA, based on different or fishing effort data are not necessary. One should assumptions about terminal F, all fit the data not, however, allow oneself to be seduced by all equally well (exactly), and cannot therefore be dis- this convenience. Regarded as a statistical proce- tinguished on the basis of their goodness-of-fit. dure, a simple trial ‘untuned’ VPA is completely There is no doubt, however, that some of the pos- underdetermined, and can at best be thought of as a sible solutions would be regarded by any reason- non-unique transformation of the data. The catch able person as more plausible than others. An data are also treated as though they were exact, and example of an implausible interpretation is given any sampling and measurement errors feed in Tables 7.8 and 7.9. This was constructed by in- through directly into the fishing mortality and serting alternately high and low estimates of the population estimates obtained, especially of the survivors, leading to values of approximately 1.0 survivors at the end of the final year, which is an and 0.2 alternately for terminal F. The oddity of undesirable feature. this interpretation is obvious even in Table 7.9, 136 Chapter 7

1.2

1.0

0.8

0.6 1 2 3 0.4 4

Fishing mortality at ages 1 to 4 0.2

0 Fig. 7.6 Fishing mortality 1989 1990 1991 1992 1993 1994 1995 estimates for implausible Year terminal F assumptions.

600000 1 2 500000 3

400000

300000

200000

100000 Population numbers at ages 1 to 3 0 1989 1990 1991 1992 1993 1994 1995 1996 Fig. 7.7 Population numbers for implausible terminal F Year assumptions. but may be more easily appreciated graphically, as more plausible interpretation of the data than Fig. in Figs 7.6 and 7.7. 7.6. Why? This shows that the exploitation pattern im- Fishing mortality is a function of both time and plied by this analysis is very ‘rough’, alternat- age – we usually write it as F(y,a). However, once ing between high and low values on adjacent fish have reached a certain age and size, they ages and cohorts, and that the positions of the mature, their growth slows down and they tend to highs and lows are reversed each year, creating behave as a group, so that fish of similar ages are a ‘checkerboard’ or diagonally banded pattern. found together, and are of similar size. Even if they This may be compared with the equivalent dia- deliberately tried to do so, fishermen would have gram derived from the JAM analysis, shown in difficulty in applying a much different fishing mor- Fig. 7.4. Here the exploitation pattern is more tality to adjacent age groups of the same stock – and or less flat, and has a similar shape in each year. there is no obvious reason why they should try to One feels instinctively that Fig. 7.4 represents a do so. One would therefore suppose that fishing Dynamic Pool Models I: Interpreting the Past 137 mortality should not change very much between simplest and most common is simply to specify adjacent age groups in the same year, at least once that S is defined to have the value 1.0 at some age, the fish are mature and fully recruited. One would the age of unit for selection, denoted by a*. All the also expect that the exploitation pattern created by other S values are then defined relative to that on

fishing should not normally vary haphazardly be- this age, and the Fs values are then conditional on tween one year and the next. Progressive changes the exploitation pattern so defined. The choice of over time are quite possible, as are sharp changes, a* is in principle completely free, but some practi- for example if and when a new larger or smaller cal considerations are discussed below. minimum mesh size is enforced. These changes should, however, be systematic rather than hap- 7.4.2 Indeterminacy of terminal F hazard. It is for these reasons that one feels that the and terminal S smoother and less structured pattern of fishing mortality of Fig. 7.4 is more plausible than that of There are many possible ways of fitting such a Fig. 7.6. The argument is based on the internal con- model to a set of catch-at-age data. A number of sistency of the interpretation – one is favouring the methods based on direct least squares fitting have interpretation in which adjacent values of fishing been proposed (Pope 1974; Doubleday 1976; Gray mortality are more or less similar, and in which the 1977), but were not very successful. The reason for exploitation pattern is fairly consistent from year this was elucidated by Pope and Shepherd (1982). to year. One can remove the effect of varying year-class This argument can easily be formalized. Let us strength in the catch-at-age data by taking ratios of seek an interpretation in which the exploitation adjacent catches for the same cohort, i.e. pattern (the relative fishing mortality on each age group), denoted by S(a), is the same each year, Cya(),,. Cy()++11 a (7.6) although it is still an unspecified function of age. The overall level of fishing mortality, denoted According to the separable model, the loga- by Fs(y), where the subscript s is used to identify rithms of these ratios should be well approximated this as the overall F, subject to the assumption of by a two-way analysis of variance based on the separability (see below), may change as required main effects, a year effect and an age effect, only. In from year to year. This constitutes a separable practice one finds that such an analysis of variance model for the behaviour of fishing mortality: can indeed explain a very large part of the variance of the log:catch ratios, leaving residuals which are

Fya(),.= Fs ()() ySa (7.5) usually quite consistent with expected sampling errors. This technique, of representing a function of The consequence of this is that one can only two variables by the product of two functions of determine, from such an analysis of variance, a set one variable, is known mathematically as separa- of values for the ratios Fs(y)/Fs(y + 1) and S(a)/S(a + tion of variables. Hence the common name, sepa- 1). One has, from the row and column means (the rable VPA, used for the method of analysis based on main effects) of the log:catch ratio matrix, just this model (Pope and Shepherd 1982). It should be enough constraints to specify and determine all noted that in equation (7.5) there is a degree of arbi- these ratio uniquely. Given catch data for t years trariness in the definition of Fs and S. One could and g age groups there are only (t - 1) column take any set of values for them, multiply all the Fs means and (g - 1) row means, and these are not all values by a constant, and divide all the S values by independent because the grand mean of both sets is the same constant, and have exactly the same set the same. One therefore has enough constraints of values for F(y,a). To remove this ambiguity it is (equations) to determine only (t + g - 3) parameters. necessary to apply some normalization constraint. The normalization constraint S(a*) = 1.0 provides There are many possible ways to do this, but the an additional equation, leaving just enough to 138 Chapter 7

determine the (t - 1) ratios of Fs and the (g - 1) ratios with conventional VPA, it is usual to require the of S only. analyst to provide assumptions for terminal F and

The solution in terms of the Fs(y) and S(a) values terminal S. This means values for Fs(t) and S(g) are themselves is therefore twofold indeterminate. required where t is the last year and g the oldest age This means mathematically that attempts to group. Notice that because S has been normalized determine these parameters directly, for example to unity, the value S(g) is relative to that on the ref- by least squares, are likely to fail, because the prob- erence age a*, whilst the value of Fs(t) is absolute, lem is singular, involving the inversion of a matrix and of approximately the same size as the actual which is rank-deficient of order two, or three if no individual fishing mortality on age a* in the last normalization constraint on S is used. The argu- year. The analyst therefore again has to assume ment above is based of course on several approxi- precisely those things he would most like to deter- mations, and is not exact. For this reason the mine, namely the actual level of fishing mortality singularity will not prove to be exact in practice, in the most recent year, and a parameter which but the system is likely to have two very small controls the overall shape of the exploitation pat- eigen-values (three if S(a*) is not fixed). The prob- tern. In spite of the strong assumption of separabil- lem is very similar to that of multiple regression ity, we are still left with an analysis which can using correlated (co-linear) explanatory variables. determine the details, but not the overall underly- The solution of such ill-determined systems is ing patterns. notoriously difficult. In practice all this means that in spite of the very strong model of separabil- 7.4.3 Practical implementation of ity, as expressed by equation (7.5), and the conse- separable VPA quent reduction in the number of parameters to be determined, to many fewer than the number of Given the indeterminacy of the solutions dis- data points, the problem of determining fishing cussed above, it may be questioned whether there mortality still has no unique solution. The num- is any point in carrying out separable VPA rather ber of undetermined parameters has in fact been than, say, the JAM analysis based on averaging, reduced from (t + g - 1), the number of terminal as discussed above. We believe that there is, for F values required for conventional VPA, to 2 for several reasons. First, separable VPA provides a separable VPA, assuming that natural mortality is simple and convenient method for generating trial known, throughout, which is certainly progress, analyses, by specifying only two numbers which but not enough. The problem is structurally represent clearly defined and stated assumptions. underdetermined. Second, the fact that the analyst is required to There are still therefore infinitely many possi- make these explicit choices forces the recognition ble interpretations based on separable VPA. It of ignorance. Third, it turns out that the surviving turns out that, because the separable model is only populations determined by separable VPA are less an approximation, these alternative solutions are sensitive to measurement errors than those of the only approximately equally good fits to the data, JAM and similar methods. Finally, separable VPA and of course none of them are exact. If one had proves to be a most useful tool for examining and exact data, it would be possible to choose among determining exploitation patterns, over several them on the basis of goodness-of-fit. In practice the fairly short periods of time required. By contrast, in differences are completely swamped by sampling the more heuristic methods such as JAM, the as- errors, and goodness-of-fit is generally useless as sumptions are less clear and less explicit. Why take a guide to the correct solution. To obtain a com- the average itself? – why not 1.2 or 1.5 or 0.6 times pletely specified solution, it is therefore necessary the average? Why average over five years or ages – to supply two additional pieces of information, why not three or 10 instead? Also, if these choices usually by means of extra assumptions. Because of are made automatically instead of explicitly, the the convergence property of VPA, and by analogy analyst may be seduced into believing that he has Dynamic Pool Models I: Interpreting the Past 139 determined something which he has actually the earlier years, the fit may be quite slack. The assumed. fishing mortalities given by equation (7.7) are of A surprisingly efficient simple algorithm for course rather smooth and well-behaved, and repre- separable VPA, invented by Pope, was described by sent some sort of averaged values. Pope and Shepherd (1982). This algorithm has been The fitting procedure itself is based on the implemented several times, and is provided as part log:catch ratios, and leaves year-class strength un- of the Centre for Environment, Fisheries and determined. Once F¢(y,a) has been determined, the Aquaculture Sciences (CEFAS) Lowestoft and correspondence between it and the catch data may, ICES VPA program suites. It can also be executed however, be used to determine year-class strength. perfectly well using a spreadsheet package on a mi- Every pair of values F¢(y,a) and C(y,a) implies a crocomputer, if required. To run separable VPA, value for the current population size (in numbers), the analyst must of course supply a complete set of using the catch equation (7.1), and each of these catch-at-age data, for a minimum of about five may be converted using conventional VPA or co- years and five age groups in practice, together with hort analysis to a value of either initial year-class data or, more likely, conventional assumptions strength or terminal population. about the level of natural mortality. The first Using this procedure one should not be sur- choice to be made is the age for unit selection. This prised that the fishing mortality estimates in the is in principle arbitrary and may be chosen freely as final year are just as noisy as those in previous it will not affect the essential results. In practice it years and include occasional odd high or low val- is recommended to choose the youngest age group ues. This is a natural consequence of the proce- that is likely to be fully exploited, which can usu- dure, since such values are the result of catch data ally be guessed adequately as that age which usu- in the final year which are somewhat inconsistent ally contributes most to the catch in numbers, or with previous data for the cohort, presumably as a the next older age group. This is advisable because result of sampling and ageing errors. These anom- it helps to make the results for the exploitation alous values are not usually carried forward into pattern relatively independent of the choice of the catch forecast or other subsequent calcula- fishing mortality, and vice versa. tions, and need not be caused for great concern, al- though they may bear further investigation since 7.4.4 Interpretation of results and they may be due to data processing or transcription errors (see commentary on measurement error in the calculation of year-class strength Chapter 13, this volume). Once values for terminal F and terminal S have An example is given in Tables 7.10 and 7.11, for also been selected, the essential results of sepa- the same North Sea cod data set, using Fs(t) = 1.0 rable VPA are a set of overall fishing mortalities and S(g) = 1.0, and these results are illustrated in Figs 7.8 and 7.9. The exploitation patterns of Fs(y), and a set of selection factors S(a), normalized to S(a*) = 1. These may immediately be used to gen- Fig. 7.9 may be compared with those of Figs 7.2 and erate a full matrix of fitted separable fishing mor- 7.3: they are clearly similar to those from the JAM talities using the underlying model equation method, which is to be expected since they are, in this instance, based on similar assumptions.

Fya¢(),.= FySas ()() (7.7) 7.4.5 Discussion The result of separable VPA is in fact an inter- pretation of the data which is as far as possible con- Separable VPA is a very convenient way of generat- sistent with this separable model, but not forced to ing exploratory VPAs, and very effective in forcing fit it exactly. The closeness with which the solu- the analyst to come to terms with ignorance con- tion conforms to the model depends on a number cerning estimates of terminal F and terminal S. It is of factors, and in some part of the matrix, such as in also a useful objective way of estimating exploita- 140 Chapter 7 tion patterns over limited periods of time, for ex- minimum periods in practice. Finally, the usual ample before and after an expected change: three (e.g. CEFAS, Lowestoft and ICES) implementa- years in the absolute minimum period of time for tions include the estimation of weighted average this, but four or five would be more preferable as terminal populations, which is a most useful fea- ture, particularly with data of poor quality. Against this, the method does not permit exter- Table 7.10 North Sea cod: fishing mortality estimates nal data such as CPUE from commercial fleets or from separable VPA. surveys to be incorporated directly, and it really ex- poses the analyst’s ignorance, rather than helping Age Year him to overcome it. Furthermore, the method is based on the assumption of a constant exploitation 1990 1991 1992 1993 1994 1995 pattern, for the total international fishery, for 1 0.211 0.193 0.221 0.082 0.128 0.180 some period of time, and some range of ages. The 2 0.977 0.835 0.931 0.898 0.774 1.044 results are not, of course, forced to fit this model 3 1.016 0.976 0.885 1.087 1.117 1.295 exactly, but it does nevertheless underpin the solu- 4 0.916 0.909 0.915 1.024 1.043 0.997 tions obtained. In the case of changes of exploita- 5 0.886 0.881 0.859 0.864 0.882 0.880 tion pattern on the youngest ages, there is little

Table 7.11 North Sea cod: population estimates from separable VPA.

Age Year

1990 1991 1992 1993 1994 1995 1996

1 68369 85590 155264 67535 147955 109885 0 2 95232 45317 57804 101949 50951 106557 75138 3 20399 29346 16099 18664 34011 19243 30710 4 7900 6045 9055 5438 5155 9112 4314 5 4555 2588 1995 2968 1599 1487 2754

1.4

1.2

1.0

0.8 1 0.6 2 3 0.4 4 Fishing mortality at ages 1 to 4 0.2

0 1989 1990 1991 1992 1993 1994 1995 Fig. 7.8 Fishing mortality esti- Year mates from separable VPA. Dynamic Pool Models I: Interpreting the Past 141

180000 1 160000 2 3 140000

120000

100000

80000

60000

40000 Population numbers at ages 1 to 3 20000 0 1989 1990 1991 1992 1993 1994 1995 1996 Fig. 7.9 Population estimates from separable VPA. Year

doubt that such changes are a priori quite likely, dent information, such as indices from a series of because of differing spatial distributions of juve- recruit surveys, is vitally important for this pur- nile fish from year to year. They are, however, pose. In the absence of such information the esti- also confounded with higher sampling errors on mates from separable VPA may, however, give younger ages. It is, indeed, not at all clear how some rough indication of recruiting year-class much larger residuals on the younger ages are due strength, but these should be handled with ex- to process errors through changes of exploitation treme caution, and alternative explanations pattern and how much to sampling errors (Sparre sought, especially if they are outside the historic and Hart, Chapter 13, this Volume). In any case, range of the data. any unexpectedly high or low catches of young On balance, we consider that separable VPA is a fish due to either process in recent years will be useful tool, if carefully applied, and preferable to interpreted by separable VPA as being due to large more heuristic methods such as those discussed or small year classes, and since no long run of above, including the JAM method. When total data exists for these recent year classes there catch-at-age data is the only thing available, it may is no prospect of deciding which is the correct indeed be the method of choice, but in this situa- interpretation. tion the risk of an incorrect interpretation is high, The estimates of newly recruiting (recent) year- and independent CPUE and/or survey data are class strengths obtained by separable VPA should greatly to be desired. Their analysis is discussed therefore be treated with appropriate scepticism. below. They are based on the application of ‘normal’ lev- els of F to the few catch data available, but may be due either to a real large or small year class, a shift of exploitation pattern, or simple sampling error. 7.5 TUNING OF VPA USING An analysis of the variability (log standard error) of CPUE AND SURVEY DATA historic fishing mortality on the age in question 7.5.1 The analysis of catchability gives a rough idea of the probable imprecision of the estimates. The true interpretation cannot be The need for additional information to resolve the deduced from catch-at-age data alone: indepen- indeterminacy of VPA has been stressed above. 142 Chapter 7

100000 High Med Low Survey 10000

1000 Fig. 7.10 North Sea cod: the ‘tuning’ problem. The method changes the terminal F value in an

Population numbers and survey index attempt to make the trend line for population numbers parallel the 100 trend line found in the survey data. 01 2 3 4 5 6High, medium and low F values are Age used.

Common forms of information suitable for this The use of CPUE data on its own to provide in- purpose are indices of abundance such as catch per dices of abundance, and hence estimates of total unit effort (CPUE) from commercial fleets or mortality, using the log:catch ratio method, i.e. research vessel surveys, or indices derived from Z(y,a) = ln[u(y,a)/u(y + l,a + l)], has a long history acoustic surveys. It will be assumed here that age (see e.g. Gulland 1983). Its use for VPA tuning is, composition information is available to go with however, relatively recent. Many methods have these data, although this is not always the case, been invented, particularly by assessment work- and that a reasonable time series of say five or more ing groups during the 1970s, when VPA became years of data are available. widespread, and as its indeterminacy became un- Such data therefore provide estimates of the rel- derstood (Ulltang 1977). The reason for the multi- ative abundance of each year class through time, plicity of methods is that there are many possible and may therefore be used in principle to choose choices to be made, even assuming that one uses the VPA interpretation in terms of absolute abun- some sort of regression method. Should one regress dance most closely in accordance with them. This F on effort or CPUE on population size? What type process is usually referred to as ‘tuning’ the VPA, of regression is appropriate – predictive, functional by analogy with the process of tuning a musical or calibration? Should the regression be forced instrument to the correct absolute pitch. The through the origin? Should a logarithmic transfor- process is essentially a synthesis of two different mation be applied? What criterion of best fit sorts of information, the estimates of relative should be used? Different answers to these ques- abundance from abundance indices, and the esti- tions all generate different methods of analysis mates of absolute abundance from VPA applied to which, given the quite substantial sampling errors catch data. In the interpretation of the former we of catch-at-age and CPUE data, can easily generate have an unknown multiplicative calibration con- results which differ from one another by amounts stant, which is the catchability, and in the latter an which are of considerable practical significance, unknown additive constant, the terminal popula- even if the differences are not statistically signifi- tion. By putting the two data sets together we can cant! This is a classic example of the need to exer- in principle determine both of these unknowns, cise more care in choosing methods for the and thus the full sequence of absolute population analysis of poor data than is required if one has size, as illustrated in Fig. 7.10. good (precise) data. However, Laurec and Shepherd Dynamic Pool Models I: Interpreting the Past 143

(1983) showed that many of the doubts and confu- for example, by fitting a regression of C/P against sions can be clarified if one takes care to treat the E. problem as an exercise in statistical modelling, Similarly, using model (7.9) one would natural- and if one regards the whole process as an exercise ly apply the same procedure to in the analysis of catchability. The reasons for this are as follows. CEª qP, (7.11b) The use of fishing effort or CPUE data relies on the assumption that fishing mortality should However, the results of these analyses would be strongly related to fishing effort, preferably not generally be the same. Firstly, use of standard proportional to it, and that CPUE should be predictive regression formulae to fit these equa- similarly related to abundance. It is perhaps tions would imply different assumptions about the not quite obvious that the constants of propor- nature of the errors, and therefore lead to different tionality are the same. If E is fishing effort, u is results. In case (7.11a), effort (E) would implicitly CPUE and q is the catchability coefficient, we may be treated as exact, and uniform additive errors write would implicitly be assumed for the ratio C/P. In case (7.11b), P would be treated as exact, with uni- F = qE (7.8) form additive errors in C/E. The effects of these different assumptions would usually be greatly in- or alternatively, creased if one permitted non-zero intercepts for the regressions, which might seem to be a normal u = q¢P, (7.9) and natural thing to do. Secondly, the effects of variations of effort and where P is the population averaged over the year-class strength on these two possible relation- year in the same way as is the CPUE, and ships are quite different. Consider a typical situa- we allow the possibility that q¢ may possibly tion where effort does not change very much, but be different from q. However, by definition, u = year-class strength does. The first model (F versus C/E, and C = FP [1 - exp(-Z)]/Z, so C = FP¯ , since [1 - effort) leads to a plot like Fig. 7.11(a), whilst the exp(-Z)]/Z is just the ubiquitous averaging factor second (CPUE versus population) leads to a plot relating the average of an exponentially declining like Fig. 7.11(b). quantity to its initial value. Thus, from equation The first plot exhibits no convincing relation- (7.9) ship between F and effort. One has a cloud of points some way away from the origin. The correlation qP¢== u CE = FPE (7.10) coefficient will be very small and probably not even statistically significant, and the usual predic- and thus q¢=F/E = q. The models (7.8) and (7.9) are tive regression coefficient will likewise be very therefore precisely equivalent, and involve the small. With no convincing evidence of any rela- same catchability coefficient, so that in principle tionship, the faint-hearted analyst could easily it should not matter which of these is used as the give up and conclude that the CPUE data are basis of the analysis. worthless. Contrast this with the situation shown However, in practice, different results may in Fig. 7.11(b). Here one sees a good spread of points arise because of fitting procedures. The basic ‘ob- tracing a fairly convincing linear relationship. The servable’ quantities may be regarded as C and E correlation coefficient will be high, with a highly (raw data), and P which are simply derived from cu- significant regression coefficient, and an intercept mulative catches by VPA. Using model (7.8), one near zero. Yet these are simply different presenta- would naturally fit the relationship tions of the same data! And the position could in fact be reversed if one had a stock where year-class CPª qE, (7.11a) strength was rather invariable, but effort had 144 Chapter 7

(a) 1.4 able to apply a logarithmic transformation to the 1.2 data, since both u and P are fundamentally non- 1.0 negative quantities, and both are probably mea- 0.8 sured with something approaching a constant 0.6 coefficient of variation, rather than with constant 0.4 additive errors. Fishing mortality 0.2 Most of this confusion can, however, be avoided 0 if we remember that the whole analysis is founded 0.2 0.4 0.6 0.8 1.0 1.2 on the presumption of a more or less constant Fishing effort catchability, which is of course given by

(b) 120000 quPCEP==()= () CPEFE= . (7.12) 100000 80000 Thus q is just the slope of a straight line joining 60000 each of the points to the origin, in both Figs 7.11(a) CPUE and 7.11(b). Given the statistical properties of 40000 C and P, mentioned above, it would be reasonable 20000 and preferable to work with the logarithm of this 0 equation, i.e. 0 10000 20000 30000 40000 50000 60000

Population size ln()qCPE= ln()- ln()- ln() . (7.13)

Fig. 7.11 (a) Relationship between fishing mortality and effort; (b) relationship between CPUE and population This has the incidental benefit of circumvent- size. The graphs illustrate the relationship implied by ing any argument as to whether one should in fact (a) equation 7.8 and (b) equation 7.11(a). try to estimate the reciprocal catchability (l/q), since after log transformation this becomes an equivalent procedure. If one is prepared to treat changed considerably during the period covered catchability as a constant, for each fleet and age, by the data. then one needs to do no more than estimate the This confusion is reduced considerably if one mean of ln(q) from applying equation (7.13) to the realizes that allowing a non-zero intercept in these data. This is equivalent to estimating and using regressions is in principle inconsistent with the the geometric mean slope of the lines joining postulated models (equations 7.11(a) and 7.11(b)). the points to the origin in either of Figs 7.11(a) and If one forces the regressions through the origins the 7.11(b). This commonsense approach was suggest- discrepancy between the results using these two ed by Laurec and Shepherd (1983), and leads alternative models is much reduced: the correla- to a VPA tuning method often known as the tion coefficients based on deviations from the Laurec–Shepherd method. origin will be similar, as will the regression coeffi- The approach implied by this procedure can, cients, which are the estimates of catchability, however, be extended. One can treat the individual though they will still not be quite the same. estimates of log catchability from equation (7.13) Even then, however, there is still scope for con- as a derived quantity, whose variation with time, siderable confusion and argument. One could population size, effort and weather can potentially make an excellent case for doing the regressions be examined, and tested against the underlying de- the other way round, which would mean using a fault null hypothesis that it should be a constant. calibration rather than a predictive regression (see This constant will be different for each fleet, al- Shepherd 1997). Furthermore, when working with though preferably not varying with age for older model equation (7.11b), it would be very reason- fish. We are therefore led to consider the statistical Dynamic Pool Models I: Interpreting the Past 145 modelling of log catchability, rather than any other 7.5.2 Ad hoc methods for tuning VPA quantity. We expect that it should be approxi- mately normally distributed and homoscedastic, This approach to the analysis of catchability leads assumptions which can be tested, and we shall, in directly to practical procedures for tuning VPA. the absence of evidence to the contrary, treat it as a Starting with a trial VPA, one iteratively computes constant, because that is the underlying model catchability, analyses it by means of regressions based on our understanding of the nature of the and/or averages, uses these estimates to predict catching process, at least for demersal fish. terminal F, and repeats the procedure until it has For operational use, however, well-automated converged. Such methods are generally referred to methods of analysis and processing are required. as ad hoc tuning methods, because they are not di- All possible variations of catchability cannot be rectly based on fitting a formal statistical model. considered, but it is very important to check for They should not, however, be confused with even trends with time. This is because, regrettably, more ad hoc approaches which have sometimes much fishing effort and therefore CPUE data been used, based on regressions of F on fishing ef- are not well standardized for possible changes of fort, possibly raised in some way to the total fish- fishing power. Such standardization is not easy ery, or correlations of some measure of biomass to do well (see e.g. Gulland 1983), and even if one with CPUE. These are fraught with problems and had done a good job of allowing for changes in not recommended. the composition of the fleet, in the engines and The ad hoc tuning method of analysis, for each gears used by the different vessels and so on, there fleet and age, proceeds as follows: past estimates of is still the problem of changes of catchability log catchability over the available time period are caused by natural shifts in the spatial distribution used to make an estimate of the current value. If of the fish, or shifts in the spatial distribution of one is assuming constant catchability, this esti- fishing effort within the strata of the sampling mate is just the average. Otherwise a regression scheme in use. against time is calculated, and used to make the Because of the irregularity described, abun- prediction. A conventional predictive regression dance indices from carefully controlled series of is appropriate here, since all the errors are in the surveys by research vessels, covering the whole catchability estimates: the explanatory or inde- distribution of the stocks, are much the preferred pendent variable, time, is known exactly. In either form of data. However, the sampling levels at- case, an estimate of the standard error of the pre- tained on surveys may not be sufficient to keep the diction is also made, either from the standard devi- level of sampling errors low enough for very reli- ation of the observations about the mean, or the able analysis, especially on the older and less abun- standard three-term formula for the standard error dant age groups. It is therefore usually necessary of a further prediction from a regression. Note that also to use and analyse data from commercial the standard error of the mean, or the two-term for- sources, in which changes of catchability are more mula for the standard error of the fitted value, are likely. not what is required here. Any prediction based on A useful practical procedure is to examine such the estimate will be fully affected by the usual data for changes of catchability with time, not just level of error in the observations, so this residual because such changes are the likely consequence error must be included via the three-term formula. of imperfect standardization of the effort data, but The estimation of these standard errors is im- because other changes, such as those due to cli- portant, because one often has several sets of abun- matic changes or changes of distribution and abun- dance indices. Each may be analysed in this way, dance, which include density-dependent changes each may be used to make a prediction, and these (Myers, Chapter 6, Volume 1) are likely also to results will invariably be in conflict with one an- show up as time trends, even if time itself is not the other. Some way of reconciling these conflicting underlying causal variable. estimates is needed, and a weighted mean, with 146 Chapter 7 weights based on the inverse variances of the indi- priate and inferior, and only of historical interest. vidual estimates, is an obvious and attractive can- Readers interested in these aspects are referred to didate. This overall estimate can in fact be shown the reports of the ICES working group on the to be the minimum variance estimate, and can be Methods of Fish Stock Assessment for the period justified as a maximum likelihood estimate under 1983 to 1988 for more details. Some computer suitable and not unreasonable assumptions. It packages may still offer these variants as options, clearly makes sense, too, to give greatest weight to but we would recommend that they be avoided. these estimates which have historically given the The ICES working group also recommended that most precise predictions. Prior estimates of the other possible variants which are still virgin quality of the various data sets based, for example, should be left intact! Simulation tests of various on the number of stations worked, the proportion methods have been carried out (Pope and Shepherd of the total catch taken, or the area covered, may in 1985; Anon. 1988; Sun 1989) and confirm that the fact be quite misleading: the acid test should be the procedures described above work about as well as actual performance based on the historic data set. anything else which was then available, and as The full VPA tuning algorithm may therefore be well as could be expected, given the inevitable im- written in pseudo-code as follows: perfections in the available data. These methods are often referred to as the Pseudo-code for ad hoc tuning of VPA Laurec–Shepherd method, where catchability is Initialize: Set terminal F = 1.0 (for oldest ages and treated as constant and its average is used, and the last year) Hybrid method when a time trend of catchability Do VPA is fitted. A Mixed method is also possible where For each age trends are fitted for some fleets but not others. The For each fleet methods may be available as options in standard For each year computer packages such as the Lowestoft and Calculate ln(q) ICES VPA packages. Next year Fit model to ln(q) (by regression) Predict terminal ln(q) (and 7.5.3 Practical aspects standard error thereof) In carrying out VPA tuning and similar analyses, a Calculate terminal partial ln(F) number of practical details have to be decided, Raise to terminal total ln(F) which may have significant effects on the results. Next fleet The most important of these are discussed below, Combine to get weighted average ter- and these issues are still relevant when more ad- minal total ln(F) vanced methods are used. Retransform Next age Repeat (VPA et seq.) until converged 7.5.3.1 Inclusion or exclusion of data Print results etc. Firstly, if one has an abundance of data, one must As mentioned above, in the past, numerous decide how much of it to use. It is not obvious that variations on this preferred procedure have been it is best to use all of it. Firstly, the data for the used. For example, a similar calculation can be car- youngest and oldest age groups may be poorly sam- ried out without using the logarithmic transfor- pled, with very high sampling errors, and at best mation, or regressing q (or ln(q)) against population may add little to the analysis, and at worst may size, or fishing effort. It is also possible to perform confuse it or even cause it to crash. Prior estimates similar analyses on CPUE data which have been of precision are not usually available, but the oc- aggregated over fleets in some way. All these vari- currence of numerous zeroes is almost always ants should now be regarded as relatively inappro- diagnostic of severe sampling problems (for more Dynamic Pool Models I: Interpreting the Past 147 on sampling see Evans and Grainger, Chapter 5, certainly regard any data set which consistently this volume). Practical computer programs have to gives log standard errors of one or more, on signifi- trap zeroes and handle them as best they can, but cant age groups, with grave suspicion – that means this is usually a rather simple-minded procedure, that the prediction is not even good to within a fac- and should not be relied upon to any great extent. tor of 3 either way, meaning that it is barely yield- Age groups with more than the occasional zero in ing even an order-of-magnitude estimate. either total or individual fleet catches should be excluded from the analysis, if necessary by delet- 7.5.3.2 F on the oldest age ing them from the data files. Secondly, there is no doubt that in fisheries A second important choice in practice is what to things change, for reasons which are not always select for F on the oldest age groups. This is, of understood. Old data may therefore no longer be course, not determined by the tuning procedure, representative of current conditions. There may and must be specified in some other way. At one have been some unrecorded changes in the manner time it was common practice to set these F values of fishing, by either commercial or research ves- arbitrarily, or by trial and error, or simply to choose sels, or in the spatial distribution of the stocks a value, and not subsequently ever change it. When because of climatic or multi-species effects, or the overall levels of F are high at around 1.0, these way in which the data is worked up may have choices are adequate: the resultant F values on changed. In general, therefore, one usually wishes other ages are little affected, because of the strong to place greater reliance on recent data than on old convergence of the VPA. At moderate and low lev- data. In practice a VPA can be carried out quite sat- els of F, however, the effects are more serious, and isfactorily on five to 10 years’ data, and data more arbitrary or careless choices of F on the oldest ages than 20 years old may be regarded as ancient his- can lead to very strange results and odd exploita- tory. A convenient way of taking account of this is tion patterns. This is in fact the key to a practical to apply a tapered weighting to the data before re- solution of the problem, which is addressed in a gression or averaging, the terminology being bor- more fundamental way by the integrated and ‘sur- rowed from spectral analysis. vivors’ methods of analysis discussed later on. Finally, it is necessary to decide whether data Bearing in mind that the results of separable for all available fleets should be included in the VPA show that the shape of the exploitation pat- analysis. In principle the method of analysis is it- tern on the older ages is undetermined, and that we self testing the quality of the various data sets, and have added nothing in the ad hoc tuning methods allowing for this as far as possible, so that all avail- to determine it, it seems reasonable to choose F on able data may indeed be used. In practice, however, the oldest age group so that the shape of the ex- there is little point in including bad data along ploitation pattern is internally consistent – for ex- with good data – it will be down-weighted, and ample, terminally flat, meaning that it levels off on merely increase the labour of the data preparation the oldest ages. This can be done by setting each F and the volume of the output with little or no bene- value to some proportion of the average over a few fit. Furthermore, although modern tuning meth- of the next younger age groups. A proportion of 1.0 ods do attempt to allow for data of varying quality, and an average over about five age groups will usu- they cannot do so perfectly. They are not com- ally achieve terminal flatness, and these are suit- pletely impervious to bad data, and may be ad- able default choices. It is, however, still very versely affected by it – particularly by outliers in important not to choose low values for F on the the data for the most recent year. It is therefore ad- oldest ages without strong reasons, since this can visable to remove poor-quality data sets once these destroy the convergence property of VPA, and have been identified. Here the provision of good di- drive one to the ever-present trivial interpretation agnostic information is most helpful (see below). of low F, and large populations, everywhere. If in No hard and fast rule can be given, but we would doubt, one should err on the side of higher Fs on the 148 Chapter 7 older ages rather than low ones. It occasionally development of useful diagnostics. This approach happens that an initial tuning analysis gives sur- has at present been taken furthest in the integrated prisingly low Fs. In such a case it would be desir- statistical methods (see Quinn and Deriso 1999). able to test the effect of increasing the oldest age Fs, by increasing the proportion above 1.0, to see if this 7.5.4 Fitting and suppressing trends removes the problem. At first sight there seems to be no reason why one should not fit trends of catchability to the data 7.5.3.3 Diagnostic information from all fleets in an ad hoc tuning analysis, and this Even if one is using an analysis package which is approach has indeed been used in practice in the fairly well automated, it is important that as much past. In the present context this leads to the so- statistical examination of the data, and the fit of called Hybrid method. However, it has been the model to it, as possible should be carried out. known for some time that indiscriminate fitting This requires that good statistical diagnostics be of trends of catchability is a dangerous procedure made available. There are, of course, a large num- (see, for example, Anon. 1983). Qualitatively it is ber of potential diagnostics which could be calcu- obvious that allowing much variation in catchabil- lated, including correlation coefficients between ity must be undesirable – the whole idea of CPUE variates, tests of normality and so on. In practice tuning rests on the assumption that CPUE is an in- the most important and useful ones, when using a dicator of abundance, and any variation of catcha- regression method, are probably the estimated bility weakens that foundation. In fact, if one slope and its standard error, the standard error of allowed catchability to vary with abundance, the estimated log catchability, and the highlight- rather than with time, CPUE could in fact become ing of possible outlying observations. It is impor- independent of abundance, when catchability is tant to know the size of the slope and its standard inversely proportional to abundance, and therefore error, so that one can judge both the practical and lose all utility as an indicator. A similar but less se- the statistical significance of the slope, when de- vere problem arises with variations with time. In ciding whether or not to fit a trend of catchability, fact, we now know that allowing for an arbitrary as discussed in Section 7.5.4 below. exponential trend of catchability with time causes The standard error of prediction of ln(q) is the linearized versions of the equations one is trying to best final arbiter of the quality of the result from a solve to become singular, meaning that they ac- particular data series, at a particular age. It controls quire a zero eigen-value, and this makes the solu- both the weight attached to each individual esti- tion indeterminate again. mate, and the overall error of the final weighted es- The conclusion from this is therefore that one timate of F, which can and should be calculated. As should not fit trends of catchability unless this is mentioned above, high prediction standard errors inescapable. Indeed, one may state quite firmly indicate problems with a data set, and should never that it is highly undesirable to fit trends to the be ignored. Similarly, highlighting of possible out- catchability for all fleets. If one does, the solution liers helps quality control of the data, as it may will be at best very poorly determined, and at worst pick up data processing errors before it is too late, effectively a random number. This rules out in and will bring unusual results to the analyst’s at- practice the Hybrid method in its pure form, which tention for further investigation. These should be applies to variable catchability on all fleets. One regarded as a minimum set of diagnostics. In prin- should in fact keep catchability constant for as ciple, the more diagnostics examined the better, many fleets as possible. Results of simulation tests but it is in practice necessary to strike a balance, (Anon. 1988; Sun 1989) show that paradoxically because of the psychological disincentive attached one may in fact obtain more precise results by fix- to the study of large volumes of output. There is, ing catchability on all fleets, even when this is an however, no doubt that there is room for further erroneous assumption, than by fitting the trends, Dynamic Pool Models I: Interpreting the Past 149 even when one only fits the trends where they error although this is not guaranteed. If this is the really exist. This is because the expected reduction case, however, including both sets of data should of bias by fitting the trends is outweighed in the improve the analysis. The surveys will control the final prediction error by the associated increase in absolute size of terminal F determined, but the variance, manifest as sensitivity to noise. commercial data will help to reduce the variability Nevertheless, it would be unreasonable to of the estimates. Secondly, it is often found that just fix catchability uncritically by using the Lau- surveys give good (precise) results for younger fish, rec–Shepherd method, for all and any data sets: one not fully recruited to the commercial fisheries, but needs to start with fixed catchability on ‘at least poor results for older fish of which relatively very one reliable fleet’ – meaning one where there is a few are caught. The surveys therefore dominate high probability of little or no trend in catchability. the results for younger fish, and the commercial But this can only be based on a prior assumption, or data dominate for old fish, where the reduction more commonly prejudice, since there is no way to of estimated variance may be substantial. There test the assumption made. Indeed, selecting differ- is therefore still some point in including even ent fleets to be the standard in this way will usually unstandardized commercial data, but it would lead to different results – all the trends are just de- be much more valuable if it were properly termined with respect to one another, and all ulti- standardized. mately depends therefore on which subset of fleets are taken as the standard. One may, however, have quite strong grounds for the choice: in general a re- search vessel survey would be preferred to a com- 7.6 THE EXTENDED mercial data series, and one where it is known that SURVIVORS METHOD (XSA) great care has been devoted to standardization of 7.6.1 Introduction gear and fishing practice would be preferred to one which is known to be conducted in a more haphaz- Fully integrated statistical methods such as those ard way. Nevertheless, this remains a potential described by Quinn and Deriso 1999 and briefly weak point of this and all similar methods of analy- outlined by Sparre and Hart (Chapter 13, this vol- sis, and if one finds a discrepancy between the re- ume) are in principle the most desirable methods sults of what should be well-standardized data to use for the analysis of catch-at-age and CPUE series, one may be left in doubt about the correct data. The models so fitted are as accurate a repre- result. sentation of the dynamics of populations as pos- This also serves to emphasize that standardiza- sible, and they can allow for errors in all measured tion of effort data remains vitally important, since quantities. They do, however, usually have a large otherwise one can never have any confidence that number of parameters to be estimated, and this a data series may be taken as a standard. Indeed, at makes them computationally relatively quite de- first sight it seems that one might as well just dis- manding, and sometimes less robust. card any data from unstandardized fleets, since one These is in fact a middle way between the would be able to fit any necessary trend of catcha- known crudities of the ad hoc tuning methods, bility for them, and they will add little or nothing and the full integrated statistical methods. This is to the determination of terminal F. This would, the ‘Survivors’ method, which was devised by however, be hasty and undesirable, for two rea- Doubleday (1981). The original implementation sons. First, survey data is (one hopes) unbiased by Rivard (1980) has, however, not been very with respect to trends of catchability, but it is often widely used, partly because it was written in the quite variable in that it has a high sampling error. APL programming language, which is not univer- Conversely, commercial data may be biased by sally available, and is quite impenetrable to the trends of catchability, but may be based on higher uninitiated, and partly because it did not always levels of sampling, and hence have lower sampling give reliable results (D. Rivard and S. Gavaris, per- 150 Chapter 7 sonal communication). The original implementa- which follows. The notation used is as follows. tion also allowed for only one set of survey/CPUE Firstly, suffix k is used to index cohorts (year data, which is a considerable disadvantage in classes). Thus practice. A development of the original method (Shepherd 1999), known as Extended Survivors k = y - a. (7.14) Analysis, overcomes these difficulties, corrects an inconsistency in the original formulation, and has Suffix i is used to index ages within a cohort, and given excellent results in simulation tests. the range of i within a summation usually runs The essential idea of these Survivors methods is from the current age a to the maximum observed that they focus on determining the surviving within the cohort, amax, where population for each cohort, and use VPA, or in prac- tice, the cohort analysis algorithm of Pope (1972), agtkmax =-min() , (7.15) for the estimation of past population abundance. Thus, they are expressed directly in terms of the (where g is as usual the greatest true age group, and survivors, the variable for which results are really t is the final year). The notation cum (short for required, avoiding the error-prone projection cumulative) is used to denote the operation of through the final year characteristic of ad hoc tun- accumulating something over all subsequent ages ing procedures. Second, they treat the catch data as within a cohort, so error-free in the calibration (VPA) phase of the cal- culation, but allow for deviations from the result- amax ing population estimates when determining cum = Â . (7.16) survivors. This reduces the number of parameters ia= to be determined considerably, and permits the use of simple iterative algorithms rather than direct Later, the notation cum¢ will be used to denote ac- search minimization methods. By treating the cumulation over all previous ages, i.e. over the catch data as exact during a non-critical phase of range i = amax, a - 1. The mnemonic symbol ECZ the calculation (the VPA), one gains a substantial denotes Exponential Cumulative Z or total mor- conceptual and computational saving. The impre- tality, i.e. cision introduced is unlikely to be severe unless the total international catch-at-age data are sub- ECZ() y, a= exp{} cum[] Z() y , a . (7.17) ject to larger errors than are the data for individual fleets. This is not impossible, but it seems not to be Similarly, ECM denotes Exponential Cumulative the usual situation. The method is therefore useful (Natural) Morality: and practical, and achieves some of the benefits of the integrated methods, without incurring the ECM() y, a= exp{} cum[] M() y , a . (7.18) major extra costs.

The symbol Pt(k) is used to denote the terminal 7.6.2 Description of the method population at the end of the final year, i.e. the sur- vivors for each cohort In this section we outline the method as formu- lated by Doubleday (1981), in a rather simplified manner, since the original description allows for a Pkt()=++ Py()max11,, a max (7.19) number of confusing complications which either were not implemented by Rivard (1980), or are where we recall that amax = min(g,t - k) from rarely used in practice. The aim is to explain the equation (7.15) and similarly basis of the method, and to prepare the ground for the account of the Extended Survivors method ykgtmax =+min() , . (7.20) Dynamic Pool Models I: Interpreting the Past 151

With this notation, Pope’s cohort analysis constant with respect to time. Now, the catch data equation are treated as exact in the VPA, and the estimates of population so obtained using equation (7.22) are re- Pya(),= exp[] Mya() , Py()++11 , a garded as the best available estimates of the un- + exp[]Mya() ,2 Cya() , (7.21) known true abundances, and will be treated as error-free estimates thereof. The survey CPUE may be rewritten for the final age group in each co- data, however, also provide, through equation (7.25), a relatively error-prone estimate of these hort (k), for which the survivors are Pt(k) same abundances, once the survey has been cali- brated by determining the reciprocal catchability. Pya(),= exp[] Mya() , Pt() k+ exp[] M() ya ,2 Cya() , . (7.22) We seek therefore to determine both the survivors and the incidental variables r(a) by minimizing the Then, applying equation (7.21) repeatedly discrepancies between the VPA estimates of popu- lation Pvpa, and those determined from the CPUE, Pest, which are Pvpa() y,, a= ECM()() y a P t k+ P c () y ,, a (7.23)

Pyarauyaest (),,.= ()( ) (7.27) where Pc(y,a) is the contribution to the population arising from the raised and accumulated catches, Since the errors in Pvpa are assumed to be small, the

amax main source of errors will be those in the CPUE Pyac(),,,=+¢¢Â ECMkaaCkaa()()+¢ ¢ data. These are assumed to be log-normal, but of aa¢= variable size s2(y,a), so that ¥-exp[]05 .Mk() +¢¢ a , a . (7.24) 2 PyaPest (),= vpa() ya , exp[] N()0 ,s () ya , . (7.28) Equation (7.23) is a simple explicit expression for the population-at-age in terms of the key variables Under the assumptions of normality and indepen- to be determined (the survivors) and some con- dence (etc.) maximum likelihood estimation of P y,a stants (including the c( ) array), which depend the parameters reduces to weighted least squares only on the data and on natural mortality which is estimation, and we therefore seek to minimize taken is taken to be known or, at least given.

The second main ingredient of the method is a 2 {}ln[]Pyaest () ,- ln[] Pvpa() ya , model for the relationship between CPUE and the S . (7.29) = ÂÂ 2 population abundance. Doubleday uses the same y a s ()ya, model as that which underlies the ad hoc tuning procedures, i.e. The magnitude of the errors s2(y,a) cannot usually be taken as known a priori, and they must usually be estimated from the data. Taking them to be con- uya(),,= qaPya()( ) (7.25) stant with respect to time, so as to avoid estimat- where u has no suffix f (for fleet) since for the ing as many parameters as we have data points, and moment we assume that there is only one fleet or substituting equation (7.27), we have survey. In practice it is convenient to reverse this 2 2 equation, and write SrauyaPyaa= ÂÂ[]ln()+ ln() ,- lnvpa() ,s () . (7.30) Pya(),,= rauya()( ) (7.26) Differentiating with respect to lnr(a) and setting where r(a) denotes the reciprocal of catchability, this to zero, and rearranging, allows ln[r(a)] to be which is assumed to be a function of age, but to be written explicitly as 152 Chapter 7

2 important technical details need to be handled Âln[]Pyauyavpa()() , , s () a ln[]ra()= a . (7.31) more carefully. Â1 s2 ()a a 7.6.3 Extended survivors analysis

Since the terms Pvpa(y,a)/u(y,a), may be regarded It is quite straightforward to extend the Survivors as individual estimates of reciprocal catchability, method to allow for multiple CPUE data sets, and this simply states that the best estimate of r(a) is a to derive a non-negative estimator for the sur- weighted geometric mean of the available esti- vivors themselves using least squares. We refer to mates of it. Doubleday then used equation (7.27) to the method so obtained (Shepherd 1999) as the estimate the population P(y,a) at all ages, and a sub- Extended Survivors Analysis (XSA for short). The tractive algorithm corresponding to VPA done basic procedure, involving iterative use of VPA, forwards in time (rather than retrospectively) to calibration of reciprocal catchability, calculation estimate the survivors corresponding to each of of estimated populations, and computation of these estimates, and combined these by a weighted weighted mean estimates of survivors, is the same, arithmetic average procedure. There are a number but there are important differences of detail. of difficulties with this procedure in practice. First, In particular, a different algorithm is used for as is well known, the subtractive forward VPA the estimation of survivors, while the VPA phase algorithm can, and often does, lead to negative of the procedure (equations 7.14 through to 7.24) estimates of survivors. These infeasible estimates remains the same. Equation (7.25) is, however, could be included in the weighted mean, since this generalized to allow for several ‘fleets’ (or indices is an arithmetic mean, but they were in practice re- of abundance): placed by zeros, and it is not clear whether these zeros were, or should be, included in the mean, or uyaf(),,= qafPya()() , , , (7.32) not. Tests with a Fortran re-implementation of the procedure showed that this was a severe problem and similarly r(a) becomes r(a,f). The least squares on a number of data sets which had been analysed procedure for determining r generalizes without without difficulty by ad hoc tuning procedures. difficulty, and the final estimate or r, correspond- Generation of infeasible negative estimates of ing to equation (7.31), is given by survivors is clearly an undesirable feature, which undermines confidence in the results produced. It 2 can in fact be overcome without difficulty by using Âln[]Pvpa()( ya , uyaf , , )s () af , a the logarithm of survivors as the estimation vari- ln[]raf() , = . Â1 s2 ()af, able, and determining it by the same least squares a procedure adopted for r(a), as described below. The (7.33) use of the subtractive algorithm is in fact incon- sistent with the least squares approach, and this is The interpretation of this as a weighted geometric the source of the problem. mean of all available estimates is unchanged. The The performance of the original Survivors analogue of equation (7.27) is method has indeed been found to be rather variable in practice. It gives plausible results on some data sets, and thoroughly implausible ones on others. It Pyaf(),,= rafuyaf()( , ,, ) . (7.34) performed very poorly in the simulation tests conducted by the ICES ‘Methods’ Working Group Each survey/CPUE datum now generates an esti- in 1988 (Anon. 1988), failing to give usable results mate of the true population, and thus an estimate on several of the more demanding data sets. The of the survivors of the appropriate cohort. If now essential process is well conceived, but some one differentiates S (as generalized for multiple Dynamic Pool Models I: Interpreting the Past 153

fleets) with respect to the logarithm of survivors, Begin iterative loop in order to obtain a non-negative estimate of that Do VPA (or cohort analysis) parameter, one can show (leaving out the indices Calculate Z, ECZ, etc. for clarity) that For each fleet and age calculate weighted mean reciprocal catchabil- ity (7.33) and variance thereof ÂÂ[]w()ln Pest-- ln P t cum Z ECF = 0 (7.35) f i Next fleet and age Adjust weights (using estimated variance of r) and thus For each fleet, age and year calculate estimated population (equation ÂÂw¢-()ln Pest cum Z 7.34) f i ln Pt = , (7.36) Next fleet, age, and year ÂÂw¢ For each cohort f i calculate weighted mean survivors (equation 7.36) where w¢=w/ECF. The interpretation of equation Next cohort (7.36) is again quite straightforward. It simply Repeat loop asserts that the best estimate of survivors is a Print results, residuals, diagnostics, etc. weighted geometric mean over all available data of the populations estimated from the CPUE/survey In fact, it turns out that, as usual, catchability on data, reduced by the estimated cumulative total the oldest age is extremely ill determined, and mortality to the end of the final year. This is a com- some restriction must be imposed. In practice it is monsense result, except that the weights are modi- usual to assume that catchability for each fleet is fied by division by ECF. This term progressively constant above a certain age, so that the values reduces the weight attached to estimates based on estimated for that age are used for all subsequent older, earlier data, in addition to any explicit down- ages, in estimating the populations. This removes weighting of old data, and reflects the reduced the need to make an ad hoc assumption about utility of older data for determining the terminal fishing mortality on the oldest ages, but may lead population. This is not surprising, as the forward to rather extreme estimates for these F values. A projection of the population is closely related to, further technical detail is that, since surveys are but much more robust than, forward VPA, which is carried out at different times of the year, and com- of course very sensitive to observation errors in old mercial fleets may catch fish mainly during one data. season or another, it is desirable to relate the CPUE/survey indices to the population estimates 7.6.4 Practical considerations at the appropriate time of year. The most conve- nient way to do this is in fact to correct the indices This completes the description of the Extended to the beginning of the year using the appropriate Survivors Analysis. The calculation of reciprocal proportion of Z. This is, however, almost an catchability and survivors is of course carried out unnecessary refinement, since these adjustments iteratively, with a new VPA at each step, so there is would otherwise be allowed for in the calibration always a current estimate of cum Z for use in equa- of the catchability values. tion (7.36). The algorithm may be summarized by A more substantial refinement is to allow a the following pseudo-code: more complicated model for the catchability on the Pseudo-code for Extended Survivors Analysis youngest recruiting ages. As discussed by Shepherd Read data (1997), there is good evidence that CPUE is not lin- Set prior weights, etc. early proportional to eventual year-class strength Set terminal Fs (e.g. to 1.0) for these youngest age groups. This is easily 154 Chapter 7 allowed for in this method, since any procedure, The results are also illustrated in Figs. 7.12 and such as a regression, which permits population to 7.13. From these it can be seen that the XSA inter- be estimated from CPUE may be substituted for the pretation differs in important details from those use of the simple weighted mean. The calibration presented before. Firstly, it indicates a moderate regression procedure described by Shepherd (1997) reduction of fishing mortality rates over the is fully compatible with this, and may be used for period, particularly on the 2-year-old fish, which any desired range of ages, provided that the oldest are among the most important age groups for this age is less than that above which one wishes to set stock. The mean F on ages 2 to 6 is in fact also esti- catchability to be constant. All the technical re- mated to have fallen from 0.93 to 0.81. Secondly, finements proposed by Shepherd (1997), such as the estimated abundance of the survivors at age 3 shrinkage to the mean, setting minimum vari- is much higher than for either of the other two ances, forecast/hindcast variance inflation, etc., credible analyses given above using the JAM and may also be incorporated without difficulty. separable VPA methods. This is because both survey indices indicate a high abundance for these 7.6.5 Example fish, and the survivors estimate is adjusted accord- ingly. Evidence-based estimates for features such An example of the results of analysis by XSA is as a secular trend in fishing mortality and the given in Tables 7.12 and 7.13 below. For this analy- abundance of survivors of individual age classes, sis the CPUE data from only two surveys (the can only be obtained with any confidence when English and Scottish Groundfish Surveys) were CPUE estimates are included in the analysis. used. 7.6.6 Discussion Table 7.12 North Sea cod: fishing mortality estimates from XSA. The Survivors methods effectively combine the better features of the separable VPA and the ad hoc Age Year tuning methods. The underlying model assumes 1990 1991 1992 1993 1994 1995 that fishing mortality is separable at the fleet level only, and the CPUE/survey data are incorporated 1 0.209 0.190 0.213 0.074 0.091 0.164 in a consistent way, while data for the final year are 2 0.982 0.835 0.930 0.857 0.679 0.613 correctly treated as subject to similar errors as 3 1.018 0.989 0.887 1.089 0.993 0.938 those for other years. The final estimates are appro- 4 0.922 0.910 0.947 1.032 1.049 0.735 priately weighted averages over all available data. 5 0.813 0.893 0.861 0.938 0.898 0.889 This is achieved using a simple but efficient itera- 6 0.901 1.029 0.932 1.134 0.960 0.875 tive algorithm, derived from a simplified but plau-

Table 7.13 North Sea cod: population estimates from XSA.

Age Year

1990 1991 1992 1993 1994 1995 1996

1 69300 86900 161000 74400 205000 120000 2 96600 46000 58800 107000 56500 154000 83700 3 20800 29600 16300 19000 37100 23500 68100 4 8000 6140 9020 5510 5240 11200 7520 5 4890 2610 2020 2870 1610 1500 4410 6 566 1770 873 701 919 536 505 Dynamic Pool Models I: Interpreting the Past 155

1.2

1.0

0.8 1 0.6 2 3 4 0.4

Fishing mortality at ages 1 to 4 0.2

0 1989 1990 1991 1992 1993 1994 1995 Fig. 7.12 Fishing mortality estimates from XSA. Year

250000 1 2 200000 3

150000

100000

50000 Population numbers at ages 1 to 3

Fig. 7.13 Population numbers 0 from XSA. 1989 1990 1991 1992 1993 1994 1995 1996 sible statistical model. The main simplifying as- with suitable low weight attached to the catch sumption used is that VPA provides, once tuned, data, would be more appropriate. the most precise estimate of population abun- Nevertheless, there are many practical situa- dance, so that total international catch data may tions where the Survivors type of method is quite be treated as exact in the VPA or calibration phase appropriate, and it often seems to provide adequate of the calculation. This seems to be a reasonable computational speed, robustness and precision of assumption for stocks where the major catches are the results. The Extended Survivors Analysis is an well sampled, but would be inappropriate where improvement over the original implementation in the catch data are poorly sampled or otherwise de- three ways, namely: fective, but one or more sets of reliable survey data • allowance for multiple CPUE/survey data sets; are nevertheless available. Such situations may • consistency and robustness of the calculation of occur, and in these cases a full integrated analysis, survivors; 156 Chapter 7

• incorporation of non-constant catchability for to predation mortality, and M1, due to other forms recruiting age groups. of natural mortality. It has performed well in simulation trials and prac- tical use, and can be recommended as a useful and M = M1 + M2 (7.37) practical tool, unless the catch-at-age data are of very poor quality relative to the CPUE/survey This idea was carried into multispecies versions of data. virtual population analysis (Helgason and Gisla- son 1979; Gislason and Helgason 1985) and cohort analysis (Pope 1979; Pope and Knights 1982). The 7.7 MULTISPECIES VIRTUAL strength of these two, essentially similar, ap- POPULATION ANALYSIS proaches, were that they showed how the existing catch-at-age data from a number of predator and We saw in section 7.3 that single-species virtual prey species could be combined to estimate preda- population analysis requires the assumption that tion mortality rate. However, it was clear that in natural mortality rate is a known value, often addition to the catch-at-age data, additional data thought to be both constant with year and with would be needed to characterize the feeding of each age. But in practice we might well expect that age of each predator. Thus these early papers led to natural mortality rate would vary by age, because the establishment of a working group of ICES, smaller fish are more vulnerable to predation than which undertook to make a comprehensive study larger fish. We might also expect levels of preda- of the feeding of the major predatory fish of the tion mortality to vary from year to year because North Sea during one year, to provide suitable numbers of fish predators are known to vary from inputs to these models. year to year. For example, if the numbers of cod in This study was first conducted in 1981 and was the North Sea doubled then we might expect as a known unofficially as ‘the International Year of first approximation that the amount of food they the Stomach’. The predator species considered in ate would double. We might also expect the preda- the study were cod, whiting, haddock, saithe and tion mortality to vary if the total amount of prey mackerel. Stomach samples from size-stratified available to the predators varied from year to year. ranges of these species were taken from all parts of For example in the North Sea, saithe eat mostly the North Sea for all quarters of 1981 because, as Norway pout and herring, hence if in a particular shown for example by Daan (1973), the stomach year there was more herring we might expect contents of species such as cod were known to vary predation mortality on Norway pout to decrease. by size by area and by season. Analysis of the con- Clearly, such changes in natural mortality would tents of the stomachs of the five predators were influence the results of VPA and so might have pro- undertaken by five designated laboratories to found effects on our perception of the numbers of ensure consistency of their interpretation. The fish in the sea. It is also likely that realizing that contents of the stomachs of each predator were natural mortality could vary would also influence analysed to provide, where possible, estimates of our perception of how the fish stock should be how much of each species was eaten; not an easy managed. task with partially digested remains. In particular, Such considerations persuaded Andersen and their feeding on species for which catch-at-age data Ursin (1977) to make a model of the North Sea were available were carefully reconstructed, as in- fisheries, which included a variable predation gested weight by prey size range. There were nine mortality. Their model worked using a forward species of particular interest as prey: these were simulation of fish stocks based upon the solution the five predators themselves and the prey species of simultaneous differential equations. Their most herring, sprat, Norway pout and sandeel, for which important innovation was to partition natural catch-at-age data were also available. Thus the mortality rate (M) into two components, M2, due data on these predators and prey species were in a Dynamic Pool Models I: Interpreting the Past 157 form that could be converted to estimate how Fyas(),,= M2()()() yas ,, * Cyas ,, Dyas ,, . (7.40) many fish, of each age, of each prey species, were eaten by each age of each predator species. These Equation (7.38) shows how we would include pre- data became available in 1983 (Daan 1983; Daan dation, if predators provided catch statistics in the 1989) and the ICES multispecies assessment work- same way that fishermen do. If this were the case ing group was then established to analyse the data the multispecies cohort analysis equation would using Multispecies Virtual Population Analysis suffice to solve the problem. However, since preda- (MSVPA). The early history of this working group tors do not provide catch statistics, we must find is summarized in Pope (1991) while more recent other ways of estimating D(y,a,s). We can do this approaches can be seen in ICES (1997). by considering the quantity d(y,A,S,s,a), the num- The actual analysis made by the working bers of prey species (s) of age (a) eaten by a particu- group was based on the Helgason and Gislason lar predator species (S) of age (A) in year (y). Thus: MSVPA model, but since the Pope multispecies cohort analysis model is easier to understand, Dyas(),,= Â Â dy() , ASas ,,, . (7.41) it is used here to explain the approach. However, all A all S the formulation of feeding relationships used by Helgason and Gislason (1979) is more robust Of course we do not know what d(y,A,S,s,a) is than that of Pope (1979) and so is adopted in the either, but we can write an equation for it in terms following. of the ration R(y,A,S) of all the predators of age A The single-species cohort equation (7.2) can be and species S in the year y, and the diet proportion modified quite simply to include a term due to DP(y,A,S,a,s) that the predator obtains from prey feeding, by all quantified predators on the prey of age a and species s. Thus: species age in question. We may call this term D(y,a,s), the numbers of prey species (s) of age dy(), ASas ,,,= Ry()( , AS , * DPy , ASas ,,, ) a eaten by all quantified predators in year y (or Wt() y,, a s (7.42) perhaps in some other portion of the year such as a quarter). where Wt(y,a,s) is the average weight of the prey in the predator’s stomach. Note that this may differ Pyas(),,= exp[] M 1() as , 2 [] Cyas() ,,+ Dyas() ,, from the weight of the prey in the sea, but can be obtained from stomach samples. + exp[]MasPy111() ,()++ , a , s . (7.38) Ration may also be written as the average popu- lation sizes of the predator, times the average Note that not only does this equation contain the annual (or quarterly) diet r(y,A,S). Average values amount of the prey species eaten by all predators in the equations are denoted by an over-bar: included on the analysis, but it also uses M1 in- stead of M. In effect it treats predation deaths as Ry(),, AS= P()() y ,, ASry ,,. AS (7.43) just another form of catch. Notice also that once we have successfully performed a multispecies co- Average population size is estimated from the hort analysis and estimated population sizes then results of VPA or MSVPA, while average ration we can estimate the predation mortality rate (M2) is provided by the results of feeding experiments. and the fishing mortality rate (F) by solving: If necessary, ration may be modified from year to year in response to the size and growth of ln[]Py()()++11 , a , s Pyas , , predators. = Zyas(),, Diet proportion DP(y,A,S,a,s) may be written in = Fyas(),,+ M12() as ,+ M() yas ,, (7.39) terms of the populations of the prey species and of the predator, and a factor called the ‘suitability’ and Suit(A,S,a,s). Suitability expresses how much a 158 Chapter 7 predator of a certain age would eat of a certain prey, compared with the proportions that were found relative to the total prey biomass available, of all in stomach samples for that year. This allows possible ages a¢ of all prey species s¢. Note that in the estimates of suitability to be corrected. The this formulation it is considered to be constant for corrected estimates of suitability may then be all years, though it may be allowed to vary between used to estimate new values of d(y,A,S,a,s), and quarters, or other divisions of the year, to reflect the process is repeated until the estimated values changing diet preferences/availabilities with of suitability converge to a stable value. An season: exactly similar procedure can be adopted using the ‘exact’ multispecies virtual population analy- DP() y, A ,,, S a s sis if required. In pseudo-code we can write this as an iterative P()()( y,, a s * Wt y ,, a s * Suit A ,,, S a s ) = . loop, as follows: Â Â P()y,, a¢¢ s * Wt() y ,, a¢¢ s * Suit() A ,,, S a¢¢ s all a¢ all s¢ Pseudo-code for multispecies VPA (7.44) Choose initial values of Suit(A,S,a,s). For all species for which catch-at-age data are available . . . This equation can be solved, if we know the num- carry out single-species cohort analysis to bers of predators and of prey, and also the suitabil- obtain estimates of P(y,a,s). ity of each prey for each predator. Population Begin iterative loop numbers will of course ultimately be obtained estimate d(y,A,S,a,s) from the multispecies cohort analysis equation. perform a multispecies cohort analysis Suitabilities, however, need to be estimated using calculate estimated feeding EF(y¢,A,S,a,s) of additional data. For prey species whose numbers predators are not known, Helgason and Gislason (1979) compare with actual feeding AF(y¢,A,S,a,s) adopted the approach of including an additional obtained from stomach samples group of ‘other prey’ whose joint biomass was con- recalculate Suit(A,S,a,s,new) = Suit(A,S,a,s, sidered, in the absence of any information, to be old)*AF(y¢,A,S,a,s)/EF(y¢,A,S,a,s). constant through time. This turned out to be a very Repeat until values of population and suitability stabilizing assumption, and moreover the biomass converge. of these prey could actually be set at any arbitrary level without affecting the model. This is because When this procedure was first used concern was the model effectively estimates the product of the raised that even if the process converged, the re- biomass and suitability for these prey and hence an sults might not be unique. Studies by Magnus and assumption of a higher biomass would be balanced Magnusson (1983) indicated mathematically suffi- by an estimation of a lower suitability. cient conditions for uniqueness which appeared to Suitabilities are estimated by first guessing be robust enough to ensure convergence in most their values for each predator age and prey age com- cases. However, these sufficient, but not neces- bination, and using these guesses in combination sary, conditions excluded the possibility that a with population sizes of predator and prey age predator could eat conspecifics of the same age. groups obtained from singles-species cohort analy- Since this phenomenon is observed from time sis. This gives initial values of d(y,A,S,a,s). A mul- to time, for example small 1-year-old cod are eaten tispecies cohort analysis run with these values by large 1-year-old cod, the possibility of non- provides new estimates of population size for each uniqueness of the above procedure cannot be predator and prey species. The proportion that the entirely ruled out. However, it has yet to be ob- various prey items that the MSVPA predicts to served in practice. form of the diet of each age of each predator species In principle it is possible to tune MSVPA to in a particular year, in our case 1981, can then be trends in CPUE or survey data, in a similar fashion Dynamic Pool Models I: Interpreting the Past 159 as described in single-species assessments, and 2.0 such results could be used for setting quotas. How- Age ever, the ICES multispecies assessments working 3 group found that for the predator species, at least, 1.5 2 there was little to be gained from calculating 1 multispecies-based quotas rather than those based 2 1.0 0 upon single-species analysis. This is because the M predation mortality on the larger species occurs mostly on prerecruitment ages. Thus, while this affects the long-term yield, it has much less impact 0.5 on short-term predictions. This is because these are typically made using estimates of the numbers 0 of prerecruitment ages obtained from the results 1976 1980 1984 1988 1992 of surveys, often using results from ages older Year class than those on which the majority of the predation has occurred. Multispecies effects might well af- Fig. 7.14 North Sea cod predation mortality rate by fect quotas set for prey species, since predation af- age, for selected years, from MSVPA. M2 (1980)–M2 fects all ages of the species, but in the case of (1992) = 0.99, equivalent to a exp(0.99) = 2.7 change in recruitment. sandeel and of Norway pout in the North Sea these were usually set on an average basis. In other areas, however, multispecies effects are considered in the setting of quotas for prey species such as estimate the population and the fishing mortality capelin, and in these cases tuning of some sort be- terms of the single species VPA models, but they comes important. For more details of the behav- also estimate predation mortality and suitability. iour and ecology of age-specific predation, see As with the single-species models there is no good- Chapters 11, 12, 13, Volume I, and Chapter 16, this ness-of-fit measure. This is because all the degrees volume). of freedom in the data are absorbed by the model. In the North Sea the more interesting features of At the time the models were developed, more the analysis relate to the changes observed in nat- restrictively parameterized, least-square-fitted ural mortality rate with age and with time, and the models would not have been practical, due to limi- consequent effects this has on recruitment esti- tations in computer power. It was, however, pos- mates. This is illustrated in Fig. 7.14, which sible to fit more restrictive models to the results shows the cumulative predation mortality of of MSVPA using simple statistical packages. This North Sea cod at age, estimated by MSVPA. The fig- was performed for both estimates of M2(y¢,A,S, ure clearly shows that predation mortality rate can a,s)/[P(y¢,A,S)*Wt(y¢,A,S)] and of Suit(A,S,a,s). vary with both year and age and that thus the as- Quite reasonable fits of both outputs can be sumption of constant natural mortality rate typi- obtained using predator–prey species interaction cally used in single-species models is violated. The terms int(S,s) and an Ursin (1973) log-normal figure indicates that the cumulative predation predator weight/prey weight food-preference mortality to age 3 varied by nearly 1.0 between a model, N{ln[Wt(A,S)/Wt(a,s)], m, s}. high value in 1980 and a low value in 1992. This is equivalent to a change in survival and hence of Suit() A,,, S a s apparent recruitment over these ages by a factor = int()Ss , * N{} ln[] WtAS ( , ) Wtas () , ,ms ,+ e . (7.45) of 2.7. Inevitably, multispecies virtual population These statistical models can absorb up to half of analysis, and multispecies cohort analysis, are the variation in estimates of suitability or of M2. heavily over-parameterized, since not only do they The remaining variation e is probably what one 160 Chapter 7 should expect from what must be rather variable 1.4 1991 data on stomach content data broken down by age 1.2 of both predator and prey. 1981 Clearly, simple MSVPA is dependent on the 1.0 assumption that suitability is constant through time. If this assumption were not made it would 0.8 not be possible to base it upon one year’s stomach 0.6 content data. Historical estimates might still be Suitability obtained if stomach content data were collected 0.4 each year and suitability estimated on an annual 0.2 basis. However, since in the North Sea the stomach content data collected in 1981 required a 0 large sampling effort and the dedicated work of five 1000 100 10 1 stomach-content analysis teams for two years, this Predator wt/prey wt would scarcely be practical. Moreover, if suitabil- ity changed from year to year it might prove dif- Fig. 7.15 Suitability of cod as prey for cod as a function of predator weight/prey weight. ficult to make forward predictions. Thus, it was considered important to test the hypothesis of con- stant suitability. This was attempted by collecting partial stomach samples in 1985, 1986 and 1987. ability of cod (as prey) for cod (as predator) by Analyses of these are given in Rice et al. (1991). ln(size ratio), estimated from the 1981 and 1991 However the partial nature of these samplings data respectively. Analyses of these data may be made the tests incomplete and it was decided to found in the 1993 and 1997 reports of the ICES repeat the full sampling in 1991. This allowed Multispecies Assessment Working Group (ICES equations of the form of (7.45) to be fitted jointly to 1994, 1997). estimates of suitability based upon the 1981 data In the North Sea the comparison between 1981 and the 1991 data. and 1991 was quite testing because the herring, For most predator species the joint model cap- which is an important prey, had gone from low tured about 40% of the variation, and a smaller but abundance in 1981 to high abundance in 1991, and significant fraction of the variation (between 6% other species abundances had also shifted. Given and 29%) could be captured by allowing suitability the problems of sampling the North Sea, it is diffi- to change with year. Thus, from a strictly statisti- cult to carry investigations of the constancy of cal point of view the null hypothesis of no-year suitability further there. Clearly, one might expect effect has been disproved. This suggests that as- suitability to be a function of size ratio, of species sumption of constant suitability is also disproved, overlap and of the behaviours of predator and prey. though one might note that an alternative explana- One might also speculate that the diet proportion tion might be a drift in stomach-sampling proce- function described by equation (7.44) might be in dures over a 10-year period. It is also worth bearing higher powers (positive switching) of prey biomass in mind the point made by ICES (1994) that rejec- or lower powers (negative switching) rather than in tion of constant suitability implies a more com- unit powers. plex model, but not a return to single-species It is rather easier to investigate changes in suit- dynamics. While the differences found were statis- ability and thus the applicability of MSVPA in tically significant, the high number of degrees of areas where there are fewer species. Work in the freedom of the fits meant that in many cases the Baltic (ICES 1999) and at Iceland and the Barents differences were not large, and not necessarily of Sea (Stefansson and Pálsson 1997; Tjelmeland and practical significance, as may be judged from Fig. Bogstad 1998) has this benefit and the challenge 7.15. This shows the differences of the fitted suit- has been taken up. The latter areas have also seen Dynamic Pool Models I: Interpreting the Past 161 developments of forward simulation models that the populations, and correspond to very small Fs. fit to data using more statistical principles. Again This is connected with the previously mentioned this is more simply achieved in areas with fewer fact (Pope 1972) that the convergence of the VPA species to consider. depends on the size of the estimated fishing mor- tality, not the true fishing mortality. There is al- ways a danger that any iterated VPA interpretation 7.8 CONCLUSIONS may drift into this trivial solution, which is why small assumed terminal F values should be There can be no doubt that VPA is a useful and avoided wherever possible. Clearly, however, this quite widely applicable technique. It has both is no help if the fishing mortality really is very virtues and vices, and some care is needed in its ap- small. This is just one example of the extreme diffi- plication to ensure that useful results rather than culty of estimating the absolute size of a lightly misleading results are obtained. However, it is exploited population, especially in the absence of least able to determine the things that one most fishery-independent data. wishes to know – the recent trends of overall The other main difficulty with VPA is that fishing mortality and stock size, and year-class natural mortality is taken to be known, and strengths. The details, such as the sequence of usually constant with time, which is not a very highs and lows of year-class strength, and the ups plausible assumption in a system where it is most- and downs of fishing mortality, especially in the ly due to predation, and the abundances of the past, emerge quite clearly: it is the underlying predators vary with time. This issue is addressed large-scale patterns which are undetermined. The by the multispecies VPA technique, but at the cost only possible conclusion is that VPA alone is not of a much greater burden of both sampling and enough. To have any reasonable chance of assess- computational effort. Thus, whilst in principle ing the true state of the population, more infor- MSVPA should be the preferred technique, in prac- mation is needed than is contained in the array tice most assessments are still conducted primari- of total catch-at-age data. Only when combined ly on a single-species basis. This is not a serious with good-quality CPUE or survey data can it be re- problem for short-term forecasting, but can lead to garded as a reliable technique. These fundamental substantial problems for long-terms assessments. problems, together with many of the more techni- These issues are discussed in detail in Chapter 8, cal difficulties discussed in this chapter, also affect this volume. other related methods of analysis, and here the Finally, it is interesting to observe that VPA, relative simplicity and transparency of the VPA being in essence an additive procedure based on ac- calculation can help, as it is easier to understand cumulated catches, can in fact be applied to data what is going on than it is for more elaborate pertaining to some fraction of the whole popula- methods of analysis. tion. Suppose one systematically had data for just In all analyses of fisheries data, including catch- half the total catches. Assuming that one started at-age data, one should beware of stocks which are with the correct total terminal F, one would start subjected to low fishing mortality, and more espe- the calculation with just half the true population, cially of arriving at such an interpretation of the and thereafter repeatedly raise this, and add in half data for stocks where F is in reality quite high. This the true catches. At all stages the estimated popu- problem arises because VPA and most related lation would be half the true population, but the analyses always have a trivial ‘solution’ corre- fishing mortalities (from equations 7.3 and 7.4) sponding to infinitesimal values of fishing mortal- would be correct. This is obviously a highly ideal- ity everywhere. This is obvious, because if one ized situation, and in practice any available partial starts with small enough terminal F values, corre- catch data would be likely to derive from some of sponding to very large surviving populations, all many fishing fleets, or from one area representing the catches will be trivially small compared with only part of the distribution of the stock, or from 162 Chapter 7 one season of the year, and would not be a fixed and Gislason, H. and Helgason, T. (1985) Species interaction constant fraction of the total. Nevertheless, it in assessment of fish stocks with special application to seems that the main errors involved in applying the North Sea. Dana 5, 1–44. Gray, D.F. (1977) An iterative derivation of fishing and VPA to partial catch data would be in the estimated natural mortality from catch and effort data giving populations, and that the fishing mortalities measurements of goodness of fit. ICES CM, 1977/F: 33. would still be a reasonable first approximation to Gulland, J.A. (1965) Estimation of mortality rates. ICES the truth. This approach has not often been applied CM 1965, Gadoid Fish Committee., No 3, ((Annex to in practice (not deliberately, at any rate), and it Arctic fisheries working group report)). would be interesting to explore the conditions Gulland, J.A. (1983) Fish Stock Assessment. Chichester, under which it could be used without incurring UK: Wiley. Helgason, T. and Gislason, H. (1979) VPA analysis with serious errors. species interaction due to predation. ICES CM, 1979/G: 52. Hilborn, R. and Walters, C.J. (1992) Quantitative Fish- REFERENCES eries Stock Assessment: Choice, Dynamics and Un- certainty. New York: Chapman & Hall. Andersen, K.P. and Ursin, E. (1977) A multispecies exten- ICES (1994) Report of the Multispecies Assessment sion to the Beverton and Holt theory of fishing, with Working Group. ICES CM, 1994/Assess: 9. accounts of the phosphorus circulation and primary ICES (1997) Report of the Multispecies Assessment production. Meddelelser fra Danmarks Fiskeri og Working Group. ICES CM, 1997/Assess: 16. Havundersogelser n.s., 7, 319–435. ICES (1999) Report of the Study Group on Multispecies Anon. (1983) Report of the working group on methods of Model Implementation in the Baltic. ICES CM, 1999 fish stock assessments (Copenhagen, 1983). ICES CM, H: 5, 1–200. 1983/Assess: 17. Copenhagen: ICES. Laurec, A. and Shepherd, J.G. (1983) On the analysis of Anon. (1988) Report of the working group on methods of catch and effort data: Journal du Conseil International fish stock assessment (Reykjavik, 1988). ICES CM, pour l’Exploration de la Mer 41, 81–4. 1988/Assess: 26. Magnus, R.J. and Magnusson, K.G. (1983) Existance Anon. (1991) Report of the North Sea roundfish working and uniqueness of solutions to the multispecies VPA group: ICES CM, 1991/Assess: 4. equations. ICES CM, 1983/D: 20. Beverton, R.J.H. and Holt, S.J. (1957) On the dynamics of Pope, J.G. (1972) An investigation of the accuracy of vir- exploited fish populations. Fishery Investigations. tual population analysis using cohort analysis. ICNAF London: HMSO, Ser 2 (19). Research Bulletin 9, 65–74. Cushing, D.H. (1983) Key Papers on Fish Populations. Pope, J.G. (1974) A possible alternative method to virtual Oxford: IRL Press. population analysis for the calculation of fishing mor- Daan, N. (1973) A quantitative analysis of the food intake tality from catch-at-age data. International Commu- of North Sea cod, Gadus morhua. Netherlands Journal nications in Northwest Atlantic Fisheries, Research of Search Research 6 (4), 479–517. Documents, no. 20 (Serial No. 3166). Daan, N. (1983) Analysis of the cod data collected during Pope, J.G. (1979) A modified cohort analysis in which the 1981 stomach sampling project. ICES CM, 1983/G: constant natural mortality is replaced by estimates of 61. predation levels. ICES CM, 1979/H: 16. Daan N. (ed.) (1989) Database report of the stomach sam- Pope, J.G. (1991) The ICES Multispecies Assessment pling project 1981. Coop. Res. Rep. Cons. int. Explor. Working Group: evolution, insights, and future prob- Mer 164. lems. ICES Marine Science Symposium 193, 22–33. Doubleday, W.G. (1976) A least squares approach to Pope, J.G. and Knights, B.J. (1982) Simple models of pre- analysing catch at age data. Res. Bull. Int. Comm. dation in multi-age multispecies fisheries for consider- Northw. Atl. Fish. 12, 69–81. ing the estimation of fishing mortality and its effects. Doubleday, W.G. (1981) A method for estimating the P64–69. In: M.C. Mercer (ed.) Multispecies Approaches abundance of survivors of an exploited fish population to Fisheries Management Advice. Can. Spec. Publ. using commercial catch-at-age and research vessel Fish. Aquat. Sci. 59, pp. 64–9. abundance indices. Special Publication, Canadian Pope, J.G. and Shepherd, J.G. (1982) A simple method Journal of Fisheries and Aquatic Sciences 58, 164–78. for the consistent interpretation of catch at age data. Fry, F.E.J. (1949) Statistics of a lake trout fishery. Journal du Conseil International pour l’Exploration Biometrics 5, 26–67. de la Mer 40 (2), 176–84. Dynamic Pool Models I: Interpreting the Past 163

Pope, J.G. and Shepherd, J.G. (1985) A comparison of the improved method for the analysis of catch-at-age performance of various methods for tuning VPAs using data and abundance indices. ICES Journal of Marine effort data. Journal du Conseil International pour Science 56, 584–91. l’Exploration de la Mer 42, 129–51. Stefansson, G. and Pálsson, O.K. (1997) A Boreal Migra- Quinn, T.J. and Deriso, R.B. (1999) Quantitative Fish tion and Consumption Model. Hafrannsóknastofnun Dynamics. Oxford: Oxford University Press. Fjölrit NR. Marine Research Institute Report, 58. Rice, J., Daan, N., Pope, J.G. and Gislason, H. (1991) The Stephenson, G. (1973) Mathematical Methods for stability of estimates of suitabilities in MSVPA, over Science Students. London: Longman. 4 years of data from predator stomachs. Journal of Sun, M. (1989) Simulation testing of the performance of Marine Science Symposium 193, 3445. methods for fish stock assessment and short-term fore- Ricker, W.E. (1975) Computation and interpretation of casts. PhD thesis, University of East Anglia. biological statistics of fish populations. Bulletin of the Tjelmeland, S. and Bogstad, B. (1998) MULTSPEC – a re- Fisheries Research Board of Canada 191, 382. view of multispecies modeling project for the Barents Rivard, D. (1980) APL programs for stock assessment. Sea. Fisheries Research 37, 127–42. Canadian Technical Reports on Fisheries and Aquatic Ulltang, O. (1977) Sources of errors in and limitations of Science 53, 103. virtual population analysis (cohort analysis). Journal Shepherd, J.G. (1997) Prediction of year-class strength by du Conseil International pour l’Exploration de la Mer calibration regression analysis of multiple recruit 37, 249–60. index series. ICES Journal of Marine Science 54, Ursin, E. (1973) On the prey size preferences of cod and 741–52. dab. Meddelelser fra Denmarks Fiskeri og Havunder- Shepherd, J.G. (1999) Extended Survivors Analysis: an sogelser N.S., 7, 95–8. 8 Dynamic Pool Models II: Short-term and Long-term Forecasts of Catch and Biomass

J.G. SHEPHERD AND J.G. POPE

The essential ingredients of a dynamic pool 8.1 SHORT-TERM short-term forecast are: FORECASTS OF CATCH 1 an estimate of the current population size; AND BIOMASS 2 an estimate of the likely exploitation pattern; 8.1.1 Introduction 3 estimates of the sizes of newly recruiting year classes; An essential task in fish stock assessment is the 4 some assumptions concerning future overall preparation of short-term forecasts of catch and levels of fishing mortality. biomass. Whenever management is based on Total Note that the first three ingredients are all Allowable Catches (TACs) and quotas, such calcu- actual estimates of the current or future situation, lations are a necessary part of the provision of whereas the last is just a set of assumptions of what management advice. Indeed, so important had this may possibly occur, rather than an estimate or pre- task become in certain areas, such as the North diction of what will actually occur. These assump- Atlantic, that it had until recently almost over- tions are usually used to generate a range of catch shadowed the other essential ingredients of options, from which managers are expected to an assessment – and regrettably had received choose an appropriate course of action. In doing higher priority than the provision of long-term so it is to be hoped that they will also have some advice, such as that based on yield-per-recruit idea of the long-term strategy which they are at- and biomass-per-recruit analysis, and the stock– tempting to implement, and some guidance on recruitment relationship, which is discussed later the probable consequences, in both the short- and in this chapter in Section 8.2. longer-term, of the options presented. In some A number of methods are used for the prepara- parts of the world the options are more constrained tion of short-term catch forecasts, depending on by management policy and strategic decisions the data available, and the type of assessment that which have already been taken, so that it may be has been carried out. The ‘classical’ dynamic pool necessary to carry out calculations for only one method, which is based on estimates of population option, for example, that corresponding to the and fishing mortality for each age group, is the sub- F0.1 level of fishing mortality. The nature of the ject of this chapter. Methods based on simpler mul- calculation is, however, the same. tiplicative models and on length compositions are The distinction between ingredients 2 and 4 is also possible, but are not treated here (see Pitcher, not very precise, because the future exploitation Chapter 9, this volume). Methods based on stock- pattern may sometimes be expected to differ from production models are discussed by Schnute and that of the recent past, if one is about to implement Richards (Chapter 6, this volume). a mesh change, for example. We have separated Dynamic Pool Models II: Forecasting 165 them in order to stress that the exploitation pat- The summation extends over all ages, from that tern can be estimated from recent data, and often of first recruitment to the fishery, up to and includ- can reasonably be assumed to remain the same for ing that of the greatest age (g), which is usually a the immediate future. The future level of overall plus group. This means that it includes fish of age fishing mortality is, however, not only uncertain, g and older simply so that no fish are ‘lost’ from and not necessarily similar to recent levels – it is, the calculation. A modified form of equation (8.2) moreover, just this which man-agement action is thus required to take account of the plus group. normally seeks to control. At the stage of preparing This is just catch option calculations, it must therefore be treated as unknown, and can only be fixed by as- Py()+ 1, g= Pyg() , exp[]- Zyg() , sumption, not from the data. +-Pyg(),11 exp[]-- Zyg() , . (8.4)

8.1.2 The basic method In practice, the two values of Z in this equation are usually taken to be the same, giving a small An age-based catch forecast is very simply carried simplification. out using two of the fundamental equations of fish The calculation defined by equation (8.3) is population dynamics (see e.g. Sparre and Hart, clearly a simple sum-of-products, which can easily Chapter 13, this volume). First, we use the catch and conveniently be carried out using a spread- equation sheet. An example of the calculation is given below, but first it is necessary to be explicit about Cya(),= FyaPya()() , ,[]1-- exp() Zya() , Zya() , the source of the estimates of F(y,a) and P(y,a) (8.1) needed as a starting point. The estimates of P(y,a) required for these calcu- where C represents catch in number of fish of age a lations are generally obtained from virtual popula- in year y, and similarly F denotes the fishing mor- tion analyses (VPA) (see Chapter 7 this volume), or tality rate per year, P the population in number at a related calculation such as an Integrated or the beginning of each year, and Z represents the Survivors analysis. They are in fact the estimates total mortality rate. Next, we also use the usual of terminal population (survivors) obtained from population evolution equation these analyses. They will therefore be averaged over the available data to some extent, at least Py()++11, a= Pya() , exp[]- Zya() , . (8.2) for separable VPA, any multi-fleet-tuned VPA, Sur- vivors analysis and most more complex methods These, of course, are the discrete forms of the (see Chapter 7, this volume). This is a Good Thing. underlying differential equations, integrated over The estimates of F(y,a) required to apply equa- a finite time step which is usually one year as tion (8.1) may be obtained in various ways. These shown here. They would therefore be exact, pro- generally involve the application of assumed vided that F and Z remained constant during the levels of future overall fishing mortality F*(y) time period in question. The total catch in weight which are usually based on estimates from the (or yield, Y) is simply calculated by summation recent past, to a specific exploitation pattern, S(a), over all the age groups: where S(a) is just the fraction of the overall F* suf- fered by fish of each age:

Yy()= CyaWa()(), c (8.3) Âa Fya(),*= F()() ySa . (8.5) where Wc represents the average weight at age of fish in the catch, which is often different to that of Methods for estimating these recent levels of those in the stock, Ws, because fishing is a biased fishing mortality, and thus the exploitation pat- size-selective process. tern, are discussed in detail in Chapter 7 (this 166 Chapter 8 volume). There are several possible ways of con- as possible before they appear in forecast calcula- structing estimates of overall fishing mortality. tions. Such suppression cannot, of course, ever be The most usual is just to take an arithmetic mean achieved perfectly, and averaging is usually just over a suitable, but not too large, range of older and the best blunt instrument available for the fully exploited age groups. purpose. A second, almost precisely equivalent, proce- It is therefore usually highly undesirable to sim- dure is to apply multiplicative scaling factors ply take the last year’s values, F(t,a) and use those known as F-multipliers (Fmult) to some reference for a forecast. One exception to this rule is if the array of fishing mortalities: Judicious Averaging Method (see Chapter 7, this volume) is being used, since in this case the F(t,a)

Fya(),.=¥ Fmult F ref () a (8.6) are already averages, and are therefore relatively stable estimates. However, this does not make this The F-multipliers are in practice usually just a a desirable method, because, as noted in Chapter 7 set of standard numbers representing a range of in- (this volume) any errors are simply transferred to creases and decreases of fishing mortality, relative the population estimates. to the most recent level, chosen to reflect various The third main ingredient, estimates of the possible options for future management of the strengths of forthcoming year classes, which have stock. The reference fishing mortalities at age not yet been observed in the catch-at-age data, is,

Fref (a) are usually obtained by averaging over the however, often among the most important. Ideally estimates for a few (typically 3 to 5) recent years, such estimates should be based on rigorous analy- sis of suitable survey indices, and methods for

Fref() a= mean y [] F() y,, a (8.7) doing this are discussed by Shepherd (1997). Where suitable survey data are not available, less satis- where the required estimates of recent values for factory methods must be applied. These boil F(y,a) are again usually obtained from some form of down to intelligent guesswork, deliberate cau- VPA (see Chapter 7, this volume). The stock bio- tious underestimation, and the use of averages, or masses are, of course, also calculated by a simple estimates from stock-recruitment analyses (see weighted sum of products Chapter 6, Volume 1). None of these has useful short-term predictive ability in practice, and none

By()= Ws () afx ()( aPya, ) (8.8) can be considered satisfactory unless recruitment Âa to the fishery is a very gradual process, and recruit- where Ws(a) is the appropriate weight-at-age of ment, expressed as year-class strength, is itself a fish in the stock, and fx(a) is the appropriate frac- rather stable quantity. The use of averages for the tion of the population to be included at each age. strengths of any year classes which will contribute This would be the fraction of mature fish at age a if substantially to the fishery during the period spawning stock biomass is required, or the fraction covered by the forecast should be considered to of fully exploited fishing mortality suffered by fish be a desperate measure, and such forecasts of age a, if exploited stock biomass is required. should be heavily qualified with suitable health In both cases, the estimates of future F(y,a) in- warnings. volve a substantial element of averaging. This is One important question in practice is how also a Good Thing, because by convention almost many year-class strengths, as estimated by VPA or all methods of analysis in current use allow any other methods, should be replaced by independent errors in the catch-at-age data to appear in the out- estimates from surveys? In principle the use of in- put estimates and tables of fishing mortality. Such tegrated methods, which combine the analysis of errors, whatever their origin, are by definition not catch-at-age and survey data for both recruited expected to be repeated systematically in the and recruiting year classes, should reduce or elimi- future: they should therefore be suppressed so far nate this problem. In practice, however, available Dynamic Pool Models II: Forecasting 167 methods for estimating recruitment from survey year classes whose size has been assumed to be data are mostly distinct from those based on VPA 50000. The overall effects (Fig. 8.3) are therefore used for analysis of catch data. A sensible rule of a progressive increase of total catch weight, and a thumb seems to be to replace, if possible, any esti- decrease of future spawning stock biomass, as the mates for year classes on which the cumulative fishing mortality is increased. This figure summa- fishing mortality to date is less than about 0.7, so rizes the information on the short-term evolution that we have seen less than half the year class in of the stock under the three selected management the catches. If one wished to be more scientific, options, expressed as Fmult = 0.75, 1.00 and 1.25, one could instead replace any estimates for which which would be considered when selecting the the estimated standard error of the terminal F, as level for a recommendation for the Total Allow- estimated from the VPA tuning procedure or else- able Catch (TAC) for the stock. where, exceeds that for the year-class strength, de- This information would also usually be sum- rived from the recruit index analysis. marized in an option table, such as Table 8.2, and The simple sums-of-products calculation de- the usual procedure would simply be to pick as the scribed here is, of course, ideal for implementation TAC the catch weight corresponding to whichever as a spreadsheet, and for the simplest form of fore- option was considered to provide the best compro- cast considered so far there is little need to seek mise between allowing continued fishing at an ac- anything else. An example of a spreadsheet imple- ceptable level, whilst still preserving an adequate mentation for three options concerning the future biomass for the future (see also Table 1.1, Chapter level of fishing mortality is given in Table 8.1. For 1, this volume). Such a selection is often the sub- simplicity and clarity, this illustrates a catch fore- ject of considerable uncertainty and controversy. cast only for a single year, although a two-year In general the assessment scientists are likely to catch forecast is usually required when an opera- take a long-term view and argue on behalf of the tional assessment is carried out. This is because fish, for a low F, high SSB option, whereas the fish- an assessment is usually carried out in year (t + 1) ing industry is likely to take a short-term eco- using data up to and including the last full year t, nomic view and argue for a high F, high-catch option. and a forecast is required of catches in the next full year (t + 2). It is also usually necessary to calculate 8.1.3 Complications the stock biomass at the end of year (t + 2), which is of course the beginning of year (t + 3), in order to A spreadsheet implementation is perfectly satis- illustrate the full effect of choosing any particular factory for such simple forecasts for a couple of catch option. The extension of the calculation to years. However, there are a number of additional more years is straightforward. factors which need to be considered in most real- The results of this forecast are also illustrated in life situations, which can lead to considerable Figs 8.1 to 8.3. These show how the catch-at-age complication of this fundamentally straightfor- increases as the forecast fishing mortality is in- ward procedure. These include, in particular: creased by increasing the F-multiplier (Fig. 8.1), • calculation of a large number of management whilst the future population number at each age options; decreases (Fig. 8.2). Note how the catches are dom- • medium or long-term forecasts extending over a inated by the strong year-class contributing at age longer time horizon; 3 in these data, which also makes a major contribu- • allowing for discards; tion to the future population at age 4. The VPA • calculations for multiple fleets; results indicate that the subsequent year class is • simultaneous calculations for several species in weak, and it makes an unusually small contribu- a mixed fishery; tion to the catches at age 2 and the population at • allowing for changes of exploitation pattern. age 3. The contributions to the catches at age 1, and Some observations on each of these aspects to the future population at ages 1 and 2, are due to follow: 168 Chapter 8

Table 8.1 Catch forecast spreadsheet.

Age Weight Maturity Current Current Recent Expected Forecast Forecast Forecast Forecast at age at age population SSB F at age F at age catch catch population SSB (from VPA) (numbers) (weight) (numbers)

M = 0.20 Fmult = 0.75 1 0.69 0.01 50000 345 0.09 0.07 2961 2043 50000 345 2 1.08 0.05 16465 889 0.64 0.48 5734 6193 38265 2066 3 2.29 0.23 63654 33527 0.83 0.62 27011 61854 8341 4393 4 4.30 0.62 12932 34477 0.81 0.61 5390 23178 27965 74555 5 6.64 0.86 5296 30242 0.76 0.57 2105 13979 5767 32934 6 8.38 1.00 800 6704 0.81 0.61 333 2794 2452 20549 7+ 9.80 1.00 283 2773 0.87 0.65 124 1218 477 4679 Total Catch 111259 SSB (new) 139521 B (old) 108957 Fmult = 1.00 0.09 3906 2695 50000 345 0.64 7129 7699 37413 2020 0.83 32982 75528 7108 3744 0.81 6594 28353 22725 60585 0.76 2587 17180 4710 26896 0.81 408 3418 2028 16993 0.87 151 1482 388 3807 Catch 136356 SSB (new) 114390 Fmult = 1.25 0.11 4831 3333 50000 345 0.80 8326 8992 36581 1975 1.04 37884 86755 6057 3190 1.01 7587 32623 18467 49232 0.95 2990 19852 3847 21966 1.01 469 3933 1677 14052 1.09 173 1696 316 3097 Catch 157185 SSB (new) 93858

Notes: F = fishing mortality; Fmult = potential F values relative to the most recent F; M = natural mortality; SSB = spawning stock biomass.

1 Multiple-option calculations need usually only Table 8.2 Catch forecast option table. be provided in one year of a multi-year forecast, F preferably the final one. Allowing for several ‘op- Option -multiplier Catch weight SSB (new) (ktonne) (ktonne) tion years’ leads to an explosion of alternatives, and these are usually difficult to understand, as A 0.75 111 140 well as tricky to organize. B 1.00 136 114 2 There is no reason why the forecast calcula- C 1.25 157 94 tion may not be continued indefinitely, pro- vided that the output is suitably organized. A suitable way of supplying the future fishing relationship (see below and Myers, Chapter 6, mortalities and recruitment estimates is, of Volume 1). Allowing for indefinitely long catch course, required. In practice recruitment in the forecasts can be quite useful because it provides unforeseeable future can only realistically be a cross-check on yield-per-recruit calculations supplied as an average or from a stock–recruitment (see also below). Dynamic Pool Models II: Forecasting 169

40000

35000

30000 Fmult = 0.75 25000 Fmult = 1.00 Fmult = 1.25 20000

15000 Number at age

10000

5000

Fig. 8.1 Forecast catch numbers: 0 F-multiplier is used to modulate 12 345 67+ fishing mortality. Age

60000

50000

Fmult = 0.75 40000 Fmult = 1.00 Fmult = 1.25 30000

20000 Population number at age 10000

0 Fig. 8.2 Forecast population 12 345 67+ number at age. Age

The specification of these future fishing mortal- 180000 ities is not trivial, because there are numerous 160000 possible scenarios which could be considered. The 140000 mortality rates associated with these are often 120000 most conveniently provided as ‘F-multipliers’ 100000 rather than absolute values, as discussed above. 80000 Experience has shown that it is thoroughly confus- 60000 Catch weight ing to use year-on-year multipliers which accumu- 40000 Spawning biomass late, and that it is preferable for all F values to be 20000 expressed as multipliers relative to a single his- Catch weight and biomass (tonnes) 0 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 toric reference F-pattern. Changes in the exploita- F-multiplier tion pattern cause further complications (see item 6 below). Fig. 8.3 Forecast catch weight and spawning stock 3 Discarding at sea of fish because they are too biomass. 170 Chapter 8 small, unmarketable, or over quota, is a fact of life (f, for fleet), as do C and F, when several fleets are (see Kaiser and Jennings, Chapter 16, this volume). considered. Sometimes it is a major factor in assessing the state 4 The extension of the calculations to several of a stock and its prognosis for the future. In princi- fleets is straightforward. One uses the partial fish- ple there is no particular difficulty in including it ing mortality F(y,a,f) and calculates the catch (or in assessment calculations, whether retrospective landings) for the fleet C(y,a,f) using (see Chapter 7 this volume) or prognostic, as here. In practice there may be some problems, because Cyaf(),,= FyafPya()() ,, , the sampling errors associated with discard data ¥-{}()1 exp[] -Zya() , Zya , (8.11) are usually high. For a forecast, one needs to specify the rate of discarding, and this is most with or without the extra term for discarding, as easily done by supplying estimates of the fraction required. Again the population equation (8.2) of fish caught which are discarded, as a function needs no change, since Z(y,a) is always taken to of age or size (Sparre and Hart, Chapter 13, this refer to the total of all mortality rates, and there- volume). To a first approximation, these fractions fore includes the sum of all the partial fishing mor- may be taken to be constant over time, and for tality rates. The symbol F(y,a) (without the extra medium/long-term forecasts this is likely to be the index f) normally refers to this total fishing mortal- best available assumption. In practice the fraction ity rate, unless otherwise specified. Equation discarded is likely to vary with a number of factors, (8.3) for the yield (catch-in-weight) by each fleet of including the size of the year class concerned. If a course becomes statistical analysis of the data were carried out to establish a satisfactory predictive model for this Yyf(),,,,= CyafW()()c af (8.12) Âa variation, then it could of course be included in the calculation. since the weight-at-age of fish in the catch (or land- The catch equation is easily modified to incor- ings) may be expected to vary from fleet to fleet, porate discards, as and these must then be summed to get the total yield.

CyaLD(),,,= FyaPya()()[]1 - fa() 5 The extension to several species caught simul- ¥-{}()1 exp[] -Zya() , Zya , (8.9) taneously in a mixed fishery is also conceptually straightforward. Multispecies forecasts which in- clude biological interactions (such as the effects of where the subscript implies those fish caught and L predation) are discussed in more detail in Section landed. The symbol f (a) denotes the fraction dis- D 8.3 below. For the technical interactions caused carded of each age group. The numbers discarded only by the occurrence of mixed fisheries, the are then, of course equations just get duplicated for each species, ac- quiring yet another suffix (s, for species) on all rele- CyaFyaPyafaDD(),,,= ()()() vant quantities. The only real complication is in ¥-{}()1 exp[] -Zya() , Zya , . (8.10) the generation of the partial fishing mortalities. We now have an array of reference fishing mortali-

The equation for the evolution of the popula- ties, Fref (a,f,s), obtained by time-averaging over a tion (8.2) needs no modification, because the usual few recent years of the historic data for F(y,a,f,s). convention is that fishing mortality (and there- The future fishing mortalities are obtained by ap- fore total mortality) includes all deaths due to plying F-multipliers for each year and for each fleet fishing, whether or not the fish are landed. Finally, which represent modulations of the fishing effort looking forward to the next paragraph, discarding exerted by those fleets in the past, so that practices often vary considerably between differ- ent fleets, so that fD may require another suffix Fyafs(),,,= Fmult() yf ,* F ref ( afs ,, ) (8.13) Dynamic Pool Models II: Forecasting 171

where Fmult(y,f) is just specified as a set of numbers For these reasons such programs are often provided representing the assumptions about the future ac- and maintained by the international bodies re- tivity of each fleet, which are not directly depen- sponsible for stock assessments, such as ICES (see dent on species (s). This is consistent with the www.ices.dk). simplifying assumption that both effort and fish- There are numerous practical details to be ing mortality vary in the same way for all species taken into account when executing catch fore- caught. This implies that there is a time-invariant casts, especially concerning the selection of the set of ‘bycatch ratios’, or species-specific selection basic data representing the state of the stock, and patterns, i.e. S(a,f,s), where the assumptions to use for its future development. In particular, we recommend that the calculations Fyafs(),,,= Fyfs() ,,* Safs() ,, , (8.14) should always be started from the best possible es- timates of the current surviving populations for where F(y,f,s*) is determined for a target or refer- each age group, as free as possible from sampling ence species s*, and errors. In practice, this means taking great care, es- pecially when using ad hoc tuning methods which

S() a,, f s= meany [] F() y ,,, a f s F() y ,,*. f s (8.15) treat the total international catch data in the final data year as exact, when projecting forward to the Thus the values of S(a,f,s) are likely to be close to end of that year, and thus to the beginning of the 1.0 only for fully exploited ages of the reference next, as required. If the data are less than adequate, species. They will usually be substantially less with coefficients of variation for important age than one for all ages of the ‘bycatch’ species. Equa- groups larger than 20%, say, it may well be worth- tion (8.14) provides an alternative, although in while switching to an Integrated or Survivors practice less convenient, way of generating the method, or even using a separable VPA (as dis- partial fishing mortalities. cussed in Chapter 7) on the last five years’ data, just 6 Things get even more complicated than this if to get a stabilized estimate of the survivors and the the exploitation pattern is permitted to change exploitation pattern for the forecast. Similarly, the with time, as, for example, when one is assessing reference F array should be calculated in a stable the consequences of a change of mesh size. Here way, by appropriate averaging over time. ‘Raw’ again, to avoid confusion, it is important that the final-year F estimates from VPA should not be used new exploitation patterns S¢(a,f,s) are expressed because they contain the full sampling errors of relative to the same reference fishing mortality ar- the catch and CPUE data. rays Fref (y,f,s*), rather than by applying multipliers Finally, great care should be taken over esti- cumulatively. mates of discard proportions and weights-at-age. Both often contain real changes with time, as well 8.1.4 Discussion as discrepancies due to poor-quality data, and care- ful scrutiny of these data is necessary, preferably Catch prediction is mathematically rather simple, by statistical modelling, such as age-by-year mul- but computationally it can become rather compli- tiplicative modelling (see Shepherd and Nicholson cated and tedious. Whilst spreadsheets are ade- 1986). This can help to avoid major errors which quate for simple situations, dedicated computer may otherwise creep unnoticed into the calcula- programs are generally preferable for operational tion. A 30% error in the weight-at-age for an im- use, not least because compiled programs are portant age group could vitiate all of the effort much more secure against the introduction of devoted to careful tuning of the VPA calculation, accidental errors. They can also be constructed to and such errors can easily arise if there are trends in provide output organized according to the layout the data, or if growth depends on year-class size or required by the subsequent users of the informa- prey abundance, and this is not recognized and tion, such as the fishery management authorities. allowed for (see Jobling, Chapter 5, Volume 1). 172 Chapter 8

some effect on recruitment. In practice, the rela- 8.2 LONG-TERM FORECASTS tionship is almost never clear, and has little predic- OF CATCH AND BIOMASS tive utility, because of the very large fluctuations 8.2.1 Introduction of recruitment, presumably due to the vagaries of weather and currents. This problem of deter- In Chapter 7 (this volume), we stressed the impor- mining the stock–recruitment relationship is dis- tance of having good-quality estimates of the size cussed in detail by Myers. Even in the unlikely of newly recruiting year classes when making event that a useful predictive relationship could be short-term catch forecasts. Unless fishing morta- found, this would not solve the problem, because lity is less than about 0.3 and the variability of re- this would predict only the expected (determinis- cruitment is also low, the accuracy of the forecast tic) size of the future year classes. These would in is largely controlled by the accuracy of the esti- practice be perturbed by other events – notably the mates of these recruitment estimates. Recruit- weather. Even if the effects of the weather were ment is, however, not usually determined until understood and quantified, this would not help us around the end of the first year of life, and it may until the happy day when our meteorological col- not be possible to actually observe it until the fol- leagues were able to provide accurate and detailed lowing year, or even later, even in a research survey weather forecasts up to a decade ahead. The pre- with a small mesh net. In temperate waters, most diction would therefore still only be of expected fisheries begin to exploit fish of two or three years average recruitment, and would not include the old, and three- and four-year-old fish often consti- inevitable perturbations, except in so far as one tute the major part of the catch. Thus, it is often could calculate confidence regions for recruit- only just possible to get the observations of recruit- ment, and make stochastic forecasts. ment which are required in time to make informed Thus, medium- and long-term forecasts are at forecasts of the catches one or two years ahead. best only forecasts of average catch and biomass, When no research survey data are available, so that where the average must be understood to be taken one must rely on catches of young fish in the fish- over hypothetical ensembles of all possible out- ery itself, the situation is even worse. comes, or over enough years, which usually means Forecasts are, however, often required for more five or more, to prevent fluctuations from domi- than two years ahead. This may, for example, be nating the result. In fact, it is common practice to necessary so as to provide advice on the time-scale use some recent average of actual recruitment as of recovery of the stock after a management mea- the estimate for those year classes for which no in- sure has been implemented. This may require a formation is available. This bold assumption can forecast extending five to 10 years into the future – perhaps be justified as the least implausible one a medium-term forecast. In addition, the selection available, but one should always be aware that it of appropriate management measures, such as the may be quite unrealistic in practice. There are in selection of a mesh size, or a target level of fishing fact at least four reasons why such a recent average mortality, usually requires the comparison of the may not be a good estimate for the future: state of the stock and the fishery under the new 1 if a stock–recruitment relationship exists, even management regime, with that which would have though it cannot be demonstrated statistically to occurred had nothing been done, after everything exist, then the expected level of recruitment will has settled down to its final state, after 10 years or change as the stock size changes; even more. 2 if climatic changes occur, as they seem to do on The provision of sensible estimates of recruit- decadal and longer time-scales, the expected level ment for such calculations is obviously a major of recruitment may also change; problem (Myers, Chapter 6, Volume 1). In princi- 3 if multispecies effects are important, recruit- ple, the state of the stock, particularly the size of ment to one stock may be modulated by changes in the spawning stock, would be expected to have the size of other stocks; Dynamic Pool Models II: Forecasting 173

4 if recruitment is determined by a stochastic argued that fish stocks are never in a steady state, process, as discussed below, this may be non- but invariably subject to substantial fluctuations, stationary, so that the mean for one period is not a and that we have no basis for assuming that an reliable estimate of that for another period. equilibrium state ever exists. Whilst true, these Nevertheless, even if these problems were un- objections are not helpful, since the calculations derstood, only in cases 1 and 3 would there be any merely produce estimates which would apply if realistic hope of allowing for them. In practice, there were a steady or equilibrium state, and do not there is usually little that can be done but to use assume that such a state actually occurs. The re- some ensemble average recruitment, and try to en- sults are in any case best presented as averages, sure that the Health Warnings which should be even if this is statistically inexact, since this attached to the results do not get forgotten. Most term carries the clear and correct implication that assessment scientists learn to attach these auto- there are fluctuations too, whilst ‘steady-state’ and matically. Their customers, however, do not, and ‘equilibrium’ unfortunately suggest the opposite. may need to be reminded at regular intervals. More recently, attention has also been focused Despite this, it may sometimes be necessary to on biomass-per-recruit estimates, in attempts to present the results of medium-term forecasts in make some allowance for the possibility of stock absolute terms such as tonnes of fish, but long- collapse. The methods for computing yield and term calculations can often better be presented in biomass-per-recruit are discussed in the following relative terms, as percentage gains or losses, either section of this chapter. This is followed by a brief compared with the past or relative to another man- discussion of the stock–recruitment problem, and agement option. This is a big help, because it auto- the risk of stock collapse, and the use of stock– matically suggests that the calculation assumes recruitment relationships in long-term forecast that ‘other things remain equal’, as it does. Fur- calculations to allow for this. thermore, since the quantities of interest are catch and biomass, which usually scale proportionally 8.2.2 Yield- and biomass-per-recruit with the level of recruitment, it means that exoge- nous changes of recruitment, such as those due to The calculation of yield-per-recruit is probably chance events, or to processes 2 and 4 above, are the best-known procedure described in the classic cancelled out at least to a first approximation. This monograph of Beverton and Holt (1957). The is not, regrettably, true for the endogenous process- method they describe is, however, adapted to the es 1 and 3, where the absolute size of the stocks computational facilities of their day, and is not affects the results in a non-linear fashion. Even so, often used today. The simplest way to compute a result expressed in percentage terms is still pro- long-term average yield and biomass is often just bably less likely to be misunderstood or misused to insert an arbitrary constant value for recruit- than one in tonnage terms. ment into one of the short-term forecast calcula- The proportionality of catch and biomass to re- tions described in Chapter 7 (this volume), and step cruitment, in the absence of non-linear processes it forward in time until the results have stabilized. such as 1 and 3 or density-dependent growth, mor- This typically requires a few more annual time tality or fecundity, was recognized by the early steps than there are age groups in the calculation workers in the field, as allowing for a major sim- (the extra ones are required to let the plus group plification of some assessment problems (see settle down properly). Whilst inelegant, the Beverton and Holt 1957). This led to the adoption computations involved are trivial by modern and popularity of yield-per-recruit type calcula- standards, and this method is often useful. If tions, and these are discussed below. These are density-dependent processes such as stock– often referred to as ‘steady-state’ or ‘equilibrium’ recruitment relationships or density-dependent calculations, and this is a source of some unpro- growth are involved, this method may well be the ductive semantic argument, since it may rightly be best, since time-stepping is just as good a way of 174 Chapter 8 reaching a steady state as any other iterative Y = CaW()c () a Âa process, and has the advantage that the intermedi- = PaFa()(){}[]1-- exp[] Za() Za() Wc () a . (8.16) ate steps are also interpretable and may be useful. Âa The time-stepping calculation is, of course, just a brute force method for carrying out a sum-of- The calculation may, of course, be carried out products integration with fairly large (usually an- for more than one fleet: F(a) would then become nual) age increments. If applied unthinkingly, this the partial fishing mortality for that fleet, and may lead to non-trivial truncation errors, but even Wc(a) the weight-at-age for the catches by that these are not often troublesome, since the main fleet. P(a) still represents the total population and purpose of such calculations is most often to com- Z(a) the total mortality on the whole stock, how- pare one calculation with that for another related ever. For constant recruitment, R, at whatever age option, so such errors tend to cancel out to some is taken to be the first in the summation, and writ- extent. In addition, the standard short-term catch ing YPR for Y/R, equation (8.16) becomes forecast procedures actually involve an implicit YPR=-exp cumZ() a Wc ()() a F a analytical integration of numbers within each Âa [] year, thus overcoming the most serious possibility ¥-{}()1 exp[] -Za() Za (8.17) for error, due to significant changes of this quantity. where the notation cum for cumulative mortality The main disadvantage of this approach is the is as defined in Chapter 7 (this volume). practical difficulty of organizing and displaying Similarly, but more simply, the biomass per the results for a wide range of fishing mortalities recruit, B/R, denoted by BPR, is and, for this reason, the calculations are in practice often carried out by distinct programs written BPR= Ws () a fx () aexp - cumZ() a (8.18) Âa [] for the purpose. Nowadays, these invariably em- ploy the discrete sum-of-products method of where f (a) as usual represents the appropriate Thompson and Bell (1934). This is very simple, and x fraction of the population, in this case usually f (a) has been well described by several authors, s for the fraction mature, to give the spawning including Ricker (1975), Gulland (1983) Hilborn stock biomass-per-recruit. and Walters (1992) and Quinn and Deriso (1999). Both these quantities may easily be calculated The main advantages of this method are that there using the following pseudo-code routine. Note is no difficulty in allowing for arbitrary growth that it is assumed that the calculation is carried patterns and/or exploitation patterns, or variation out for a range of overall fishing mortalities, F* of natural mortality or discarding practices with ref so that F(a) = F* S(a), where S(a) as usual denotes age, within the same computationally straightfor- ref the exploitation pattern. ward framework. A brief account of the method is given below. The calculations may easily be imple- Pseudo-code for yield and biomass-per-recruit mented using spreadsheet software if desired, and For each value of F*ref we give an example of this. Initialize: (P = 1, YPR = 0, BPR = 0) We consider the contributions to yield and bio- For each age a mass of each age group of a year class as it passes F(a) = F*ref*S(a) through the fishery. This is equivalent to an inte- Z(a) = F(a) + M(a) gration of the catch equations with annual age BPR = BPR + P*ws(a)*fs(a) increments. The calculations may easily be gener- YPR = YPR + P*F(a)*wc(a)*Ave[Z(a)] alized to permit quarterly or monthly increments P = P*exp[-Z(a)] if so desired. Next age From equation 8.3, suppressing the variation Add contribution from the plus-group (g) with time (suffix y), the yield is BPR = BPR + P*ws(g)*fs(g)*{1 - exp[-Z(g)]} Dynamic Pool Models II: Forecasting 175

YPR = YPR + P*F(g)*wc(g)/Z(g) ished returns. The concept of F0.1, introduced by Next F*ref Gulland (see Gulland 1983), is therefore based on End an arbitrary but commonsense rule for determin- ing when further increases of F lead to little extra A spreadsheet implementation of this calcula- yield, and are likely to be uneconomic. tion is shown in Table 8.3, for a simple case with The calculation of F and F can be carried constant M, a single fleet, and no discards, for just max 0.1 out in various ways, including numerical solution one level of future fishing mortality. of the equation for the slope of the curve. This The results of a more complex case, with age- is not always straightforward and such elaboration dependent M, allowing for discards, and with re- is hardly necessary, since the simplest method sults computed for a substantial range of F to allow works perfectly well. This is to calculate YPR for the plotting of a yield-per-recruit curve, is shown about one hundred increments of F, between zero in Table 8.4. and some maximum value such as 1 or 2, and then These results are illustrated in Figs 8.4 and 8.5. to select those values of F at which the increment Also shown in Fig. 8.4 are the locations of the bio- of YPR first becomes negative, or less than 10% of logical reference points known as F and F , max 0.1 its initial value, respectively. whose use is discussed further in section 8.2.4 below. Fmax is the value of overall F for which the yield-per-recruit is a maximum. It may be very 8.2.3 Incorporating the large, or even infinite if the curve is asymptotic stock–recruitment relationship rather than domed. F , on the other hand, is a 0.1 8.2.3.1 Recruitment failure and stock collapse measure of the overall F at which a high value of yield-per-recruit can be obtained without exces- Calculations of yield and biomass per recruit are sive effort. It is determined by the point on the easy to carry out, and need only very limited data. curve at which its slope is reduced to 0.1 of that es- These are a growth curve, some assumption about timated at the origin. Thus, at F0.1, extra incre- natural mortality, and some estimates for the ments of effort produce only one-tenth the extra exploitation pattern and the maturity give. They yield that they would if the stock were very lightly have therefore been very popular for many fished, and one is deeply into a regime of dimin- decades. Their virtue is that they cancel out the

Table 8.3 North Sea cod: yield and biomass per recruit analysis. Detailed calculation for a single level of overall fishing mortality.

Age Weight F-at-age Maturity Population Biomass Catch Catch number weight

M = 0.2 1 0.69 0.09 0.01 1000 7 78 54 2 1.08 0.64 0.05 748 40 324 350 3 2.29 0.83 0.23 323 170 167 383 4 4.30 0.81 0.62 115 307 59 253 5 6.64 0.76 0.86 42 240 21 136 6 8.38 0.81 1.00 16 135 8 69 7 9.80 0.87 1.00 6 57 3 31 Total 957 1276 nb: Divide by assumed number of recruits BPR = 0.957 YPR = 1.276 Both expressed as kg/recruit 176 Chapter 8

Table 8.4 Yield and biomass per recruit: North Sea cod (with discards).

Data Age

12345678

Proportion mature 0.01 0.05 0.23 0.62 0.86 1.00 1.00 1.00 Natural mortality 0.80 0.35 0.25 0.20 0.20 0.20 0.20 0.20 Fishing mortality: human 0.14 0.76 0.92 0.78 0.73 0.76 0.78 0.76 consumption Fishing mortality: discards 0.60 0.06 0.00 0.00 0.00 0.00 0.00 0.00 Fishing mortality: total 0.74 0.82 0.92 0.78 0.73 0.76 0.78 0.76 Proportion retained 0.19 0.93 1.00 1.00 1.00 1.00 1.00 1.00 Weight-at-age: in stock 0.63 0.93 1.93 3.63 5.94 8.10 9.92 11.31 Weight-at-age: in catch 0.63 0.93 1.93 3.63 5.94 8.10 9.92 11.31 Weight-at-age: effective 0.01 0.05 0.44 2.25 5.11 8.10 9.92 11.31

Age12345678 12345678 F Stock number at age BPR Landed weight at age YPR

0.00 1.00 0.45 0.32 0.25 0.20 0.17 0.14 0.61 11.35 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.05 1.00 0.43 0.29 0.22 0.17 0.14 0.11 0.39 8.15 0.00 0.01 0.02 0.03 0.03 0.04 0.04 0.15 0.33 0.10 1.00 0.42 0.27 0.19 0.15 0.11 0.08 0.26 6.02 0.01 0.02 0.04 0.05 0.06 0.06 0.06 0.20 0.49 0.15 1.00 0.40 0.25 0.17 0.12 0.09 0.07 0.18 4.56 0.01 0.03 0.06 0.06 0.07 0.07 0.07 0.20 0.57 0.20 1.00 0.39 0.23 0.15 0.11 0.07 0.05 0.12 3.51 0.01 0.04 0.07 0.07 0.08 0.08 0.07 0.18 0.60 0.25 1.00 0.37 0.21 0.13 0.09 0.06 0.04 0.09 2.74 0.01 0.05 0.08 0.08 0.08 0.08 0.07 0.15 0.60 0.30 1.00 0.36 0.20 0.12 0.08 0.05 0.03 0.06 2.17 0.02 0.06 0.08 0.08 0.08 0.08 0.06 0.13 0.59 0.35 1.00 0.35 0.18 0.10 0.06 0.04 0.03 0.04 1.73 0.02 0.06 0.09 0.08 0.08 0.07 0.06 0.10 0.57 0.40 1.00 0.33 0.17 0.09 0.05 0.03 0.02 0.03 1.40 0.02 0.07 0.09 0.08 0.08 0.07 0.05 0.08 0.54 0.45 1.00 0.32 0.16 0.08 0.05 0.03 0.02 0.02 1.14 0.02 0.07 0.09 0.08 0.07 0.06 0.04 0.07 0.51 0.50 1.00 0.31 0.15 0.07 0.04 0.02 0.01 0.02 0.93 0.03 0.08 0.09 0.08 0.07 0.05 0.04 0.05 0.49 0.55 1.00 0.30 0.13 0.06 0.03 0.02 0.01 0.01 0.77 0.03 0.08 0.09 0.07 0.06 0.05 0.03 0.04 0.46 0.60 1.00 0.29 0.12 0.06 0.03 0.02 0.01 0.01 0.64 0.03 0.08 0.09 0.07 0.06 0.04 0.03 0.03 0.43 0.65 1.00 0.28 0.11 0.05 0.02 0.01 0.01 0.01 0.53 0.03 0.09 0.09 0.07 0.05 0.04 0.02 0.02 0.41 0.70 1.00 0.27 0.11 0.04 0.02 0.01 0.00 0.00 0.45 0.03 0.09 0.09 0.06 0.05 0.03 0.02 0.02 0.39 0.75 1.00 0.26 0.10 0.04 0.02 0.01 0.00 0.00 0.38 0.04 0.09 0.09 0.06 0.04 0.03 0.02 0.01 0.37 0.80 1.00 0.25 0.09 0.03 0.01 0.01 0.00 0.00 0.32 0.04 0.09 0.08 0.05 0.04 0.02 0.01 0.01 0.35 0.85 1.00 0.24 0.08 0.03 0.01 0.01 0.00 0.00 0.28 0.04 0.09 0.08 0.05 0.03 0.02 0.01 0.01 0.33 0.90 1.00 0.23 0.08 0.03 0.01 0.00 0.00 0.00 0.24 0.04 0.09 0.08 0.05 0.03 0.02 0.01 0.01 0.32 0.95 1.00 0.22 0.07 0.02 0.01 0.00 0.00 0.00 0.20 0.04 0.09 0.07 0.04 0.03 0.01 0.01 0.01 0.30 1.00 1.00 0.21 0.07 0.02 0.01 0.00 0.00 0.00 0.18 0.04 0.09 0.07 0.04 0.02 0.01 0.01 0.00 0.29

biggest unknown factor, recruitment, and this is however high a value is taken for fishing mortality. also their vice, since they thereby suppress what In real life, however, recruitment failure is always may be the most important factor to consider. This possible (see Myers, Chapter 6, Volume 1). There may be easily appreciated, since it would actually are a number of stocks that have collapsed at vari- be impossible ever to fish a stock that had constant ous times under heavy fishing, for example the recruitment to the point of collapse, since in this North Sea cod (Gadus morhua), herring (Clupea case neither YPR nor BPR ever reduces to zero, harengus) and mackerel, the Atlanto-Scandian Dynamic Pool Models II: Forecasting 177

1.2 tional fisheries management. The extra ingredient which produces the possibility of stock collapse is, 1.0 of course, the likelihood of reduced recruitment at reduced stock size, i.e. the stock–recruitment 0.8 relationship (SRR). Whether or not such relationships exist is a 0.6 matter of controversy (see Myers, Chapter 6, Vol- ume 1). Recruitment is usually subject to large Yield per recruit 0.4 fluctuations from year to year, and the relationship 0.2 is thus obscured. Generally the size of the spawn- F0.1 Fmax ing stock is not a powerful determinant of recruit- 0 ment. Indeed, it may not even be the major factor. 0.2 0.4 0.6 0.8 1.0 1.2 However, this is one of those cases where failure Fishing mortality to demonstrate an effect statistically does not Fig. 8.4 North Sea cod: yield per recruit. provide a licence to ignore it, because the conse- quences of doing so are too great.

12 8.2.3.2 Fitting stock–recruitment 10 relationships What sort of relationship would it be appropriate 8 to seek in a data set on spawning stock and recruit- 6 ment? In the absence of processes serving to regu- late the population size the expected relationship 4 between recruitment and parent stock size would Biomass per recruit be strict proportionality – a straight line through 2 the origin on a stock–recruitment plot. This would be the result of constant fecundity and mortality 0 rates, and this should in principle be the most 0.2 0.4 0.6 0.8 1.0 1.2 appropriate null hypothesis, to be adopted in the Fishing mortality absence of evidence to contradict it. Fig. 8.5 North Sea cod: biomass per recruit. The consequence of such an assumption is, however, that the stock would tend exponentially to either zero or infinite size, with a characteristic herring and the Canadian northern cod. In each time-scale of a few years, depending on the level of case the reduced stock size was probably only a total mortality on the stock, unless by chance this contributory factor, since most disasters are happened to be just sufficient to keep the stock caused by a conjunction of two or three factors, not in a state of neutral equilibrium. Simple graphical just one. In addition, however, one should take simulations (see e.g. Beverton and Holt 1957, Fig. note of the virtual elimination of large species 6.1) can easily be used to illustrate such behaviour, such as halibut, turbot and ling from heavily fished and the results are quite unlike anything normally areas such as the North Sea. observed in practice. It is for this reason that It is therefore imprudent to work with models almost all fishery biologists believe that some sort which neglect the possibility of stock collapse of regulatory processes are at work, even though through recruitment failure, and one must deplore they cannot be described in any detail. the excessive attention to yield-per-recruit, and If the relationship between recruitment and the Fmax and F0.1 reference points, in much tradi- parent stock is non-linear, then the possibility of 178 Chapter 8 regulation towards one or more stationary states constant recruitment at any stock level cannot exists. These may, however, be either stable or be collapsed, however great a fishing mortality is unstable equilibria, and are unlikely to be clearly applied. Some relationship between R and SSB observable in the presence of large variability. For should always be considered as soon as sufficient single-cohort spawning stocks, these occur data has been acquired, using the techniques dis- wherever the stock–recruit curve intersects a line cussed by Myers (Chapter 6, Volume 1). In practice of unit slope which passes through the origin on a the most common and useful stock–recruitment plot of recruits against spawning numbers (Ricker relationships are those proposed by: 1975). The generalization of this to multiple-age (a) Beverton and Holt (1957) spawning populations is straightforward: one sim- ply calculates the steady-state level of spawning RaBBK=+()1 (8.20) stock biomass per recruit for any desired regime of natural and fishing mortality, growth and matu- (b) Ricker (1975) rity at age, and plots the line of the corresponding slope through the origin on the diagram of recruit- RaB=-exp() BK (8.21) ment against spawning stock biomass. The inter- sections of these survival lines with the curve (c) Cushing (1973) defining recruitment as a function of spawning stock size again define potential equilibrium RaBBK= ()--()1 bb= aKBK() (8.22) states. In fact, for a stationary state, the ratio of recruitment (R) to spawning stock biomass (SSB) (d) Shepherd (1982) must equal the reciprocal of the spawning biomass-per-recruit (BPR) so that RaB=+[]1 () BKb (8.23)

R SSB= 1 BPR (8.19) (e) Schnute (1985)

This relationship is the basis for combining a RaBBK=+[]1 ()b (8.24) stock–recruitment relationship with yield- and biomass-per-recruit calculations to produce total As explained by Shepherd (1982), the three- yield curves (see e.g. Shepherd 1982, and Section parameter expressions such as (8.23) and (8.24), 8.2.3.3 below). Now, biomass-per-recruit is a mea- which are very similar in practice, can be used sure of survival and decreases as total mortality to approximate the effect of the three more increases. The equilibrium stock size therefore traditional relationships. With b=1 they are decreases as F increases, as illustrated in Fig. 8.5. It identical to the Beverton–Holt form. With b>1 is easy to see that the stock will collapse when fish- they produce a domed relationship similar to a ing mortality is so high that 1/BPR exceeds the Ricker curve. With b<1 they lead to a non- slope of the recruitment curve at the origin. This asymptotic relationship similar to that used by slope at the origin is therefore a quantity of great Cushing, with the bonus that they avoid the concern to fishery biologists. Regrettably it is not infinite slope at the origin which is the major de- guaranteed to be positive definite. It could be zero, fect of that form, since it precludes the possibility or effectively infinite, and techniques for deter- of stock collapse. Some of these relationships mining this slope for real data sets are in their are illustrated in Figs 8.6 and 8.7. Note that infancy. equations 8.20 to 8.24 have been written in a In practice there has been a tendency among dimensionally consistent form: a has the dimen- fishery biologists to take constant recruitment as sions of recruits per unit biomass, and represents the null hypothesis. As discussed above, this is the slope at the origin (except for the Cushing dangerous, because a stock capable of generating form). B and K are biomasses, and K is a threshold Dynamic Pool Models II: Forecasting 179

1.2 Beverton–Holt Ricker 1.0 Cushing

0.8

0.6 Recruitment 0.4

0.2

0 0.5 1.0 1.5 2.0 2.5 Fig. 8.6 Stock–recruitment relationships. Spawning stock biomass

1.6 Beta = 0.5 1.4 Beta = 1 Beta = 2 1.2

1.0

0.8

Recruitment 0.6

0.4

0.2

0 0.5 1.0 1.5 2.0 2.5 Fig. 8.7 Shepherd-type stock– recruitment relationships. Spawning stock biomass

biomass above which the relationship departs con- 8.2.3.3 Total yield curves and MSY siderably from linearity (again, except for 8.22). The exponent b is a dimensionless number of order The calculation of total yield and biomass curves 1, and is a shape parameter representing the degree in absolute units such as tonnes, as functions of of curvature. fishing mortality, is straightforward, given a suit- 180 Chapter 8 able SRR and previous estimates of yield-per- taken to mean that the stock should in fact have recruit (YPR) and biomass-per-recruit (BPR) as collapsed already. This can occur because these are functions of F. The procedure is as follows, using steady-state analyses, whilst real stocks are never the Shepherd form of the relationship: in a steady state. It can only be interpreted as a warning, although it does make the conclusions Pseudo-code for total yield and biomass difficult to explain when there are still fish in the calculations sea. In addition, the results depend crucially on the For each value of F parameters estimated for the stock–recruitment Calculate YPR and BPR relationship. These are always uncertain, and Calculate B = K(a*BPR - 1)1/b usually contentious. The utility of these calcula- Calculate R = B/BPR tions is in practice therefore limited, but an Calculate Y = R*YPR excellent example (Cook et al. 1997) of their use Next F predicted the demise of the North Sea cod, which is This calculation is easily integrated with that of occurring a few years later. YPR and BPR (Section 8.2.2) if desired. Several important conclusions can be drawn The results may be plotted as graphs of Y and B from such calculations: against F, or as the ‘general production’ curve of Y 1 The possibility of stock collapse is made explic- against B, as illustrated in Figs 8.8 and 8.9. These it, and some estimate of the level of fishing mortal- curves are, of course, only valid for a hypothetical ity at which it may occur is provided. steady state which never exists in practice. They 2 An estimate of Maximum Sustainable Yield represent the ensemble or time-averaged yield and (MSY) may be made, in absolute terms of tonnes of biomass that would be obtained if fishing mortali- fish, rather than as yield-per-recruit. This is nor- ty were held constant for a long time and, of course, mally less than the product of maximum yield-per- if the fitted stock–recruitment relationship were recruit and average recruitment, and may be used valid. Nevertheless, they can be quite instructive, to dampen unrealistic expectations, particularly because they do allow for the possibility of stock in the early years of a developing fishery. collapse. It is by no means uncommon for these 3 An estimate of Fmsy, the fishing mortality at curves to indicate that at current levels of F the which MSY would be attained, may also be made. stock will collapse, and this may sometimes be This is almost always less than Fmax, the fishing

0.7

0.7 0.6 0.6 MSY 0.5 0.5 0.4 0.4 0.3 0.3 0.2 Yield (absolute units) 0.2 Yield (absolute units) 0.1 0.1 Fmsy 0 0 2.0 4.0 6.0 8.0 10.0 12.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Spawning stock biomass Fishing mortality Fig. 8.9 Total yield versus spawning stock biomass Fig. 8.8 Total yield versus F (illustrative). (illustrative). Dynamic Pool Models II: Forecasting 181

mortality yielding maximum yield-per-recruit, point can be calculated, i.e. Fmsy. Since recruit- since recognizing the possibility of stock collapse ment is now specified rather than normalized out leads to a more conservative interpretation. of the calculations, the associated reference points

Taken with appropriate levels of enlightened MSY and Bmsy can also be calculated. These are, scepticism, these estimates may be useful addi- of course, the maximum sustainable yield and tional guides to the appropriateness of various the spawning stock biomass at which this would levels of exploitation. They still neglect economic be achieved. Because of the uncertainty of the factors (Chapter 12, this volume), but include stock–recruitment relationship these reference more of the relevant biology. As is often the case, points are not widely accepted, but may be useful when one includes important but uncertain fac- for convincing managers or fishermen that recent tors in an assessment, one hopes that it becomes yields have been unusually high, and that the nor- less biased, but it is likely also to become more mal expectation is much less, or that stock size has variable. fallen way below desirable levels.

Another point of interest is that the end product Fmsy is analogous to Fmax, and one can also calcu- of these calculations are production curves similar late reference points indicating maximum eco- to those obtainable from stock-production models nomic yield (MEY), provided that reliable cost and

(Schnute and Richards, Chapter 6, this volume). price information is available – F0.1 is of course a The gulf between age-based and bulked stock- surrogate for these. The actual physical or mone- production models may therefore be crossed with- tary yield (MEY) and the associated biomass out undue difficulty. Where age-based data are Bmey can also be computed together with Fmey itself available, the results of the two approaches may be (see Hannesson, Chapter 12, this volume for compared. The important difference is that the further detail). stock–recruitment relationship, with all its un- These biological reference points are some- certainties, is made explicit in the age-based times regarded as targets for management, and approach, while in the bulked-stock methods it were indeed originally intended for this purpose. lies implicit, usually unrecognized and sometimes The rationale for this is dubious, however, since implausible. Treating it explicitly should make maximum physical yield as tonnes of fish, and one aware of the limitations of the approach, and maximum economic return to a hypothetical particularly of the wide confidence limits on the monopolistic owner of the fishery, are not usually estimates, and thus engender a more realistic and the real objectives of any of the interested parties. balanced appreciation of the conclusions. The real objectives are more usually sustainable positive returns to individual fishermen, the 8.2.4 Biological reference points maintenance of employment, the utilization of existing fishing vessels, and the avoidance of social Biological reference points are values of quantities and political unrest (see also Hart and Reynolds, which are indicative of the state or character of a Chapter 1, this volume). Some further shortcom- stock, and which may be used as ‘signposts’ to ings of these reference points as objectives are dis- guide the management of the stock. The tradi- cussed below. In the light of these reservations, tional reference points Fmax and F0.1 have been they are best regarded only as signposts, or naviga- discussed in Section 8.2.2. These are indicators tion markers, which highlight the consequences of of levels of exploitation at which maximum phy- moving in one direction or another. sical yield, or something close to it, could be Fmax and F0.1 have been quite widely used as achieved under the assumption that recruitment reference points and targets of management, in is not affected by stock size. the northeast and northwest Atlantic respectively, If a stock–recruitment relationship is estimated but they suffer from the serious defect that they and allowed for, either explicitly or through a exclude the possibility of stock collapse. Whilst stock-production model, an equivalent reference this may be overcome by fitting a stock–recruitment 182 Chapter 8 curve, as described above, this is invariably a some extent on the history of the stock. If it has contentious procedure. Since the most important never been heavily exploited, yet there is indeed quantity to know is perhaps the maximum fishing some compensatory stock–recruitment relation- mortality which the stock can withstand before ship, the estimates will be too low simply because being driven to collapse, it is useful to try to do this the region where compensation is important (so directly. A procedure proposed by Shepherd (1982) that R/SSB is increased) will never have been ex- may be useful for this purpose. The fishing mortal- plored. Conversely, if a stock has been on the verge ity corresponding to the upper 10th percentile of of collapse for some time, so that one is seeing the the slopes (R/SSB) given by all the data points on a straight-line relationship between R and SSB near stock recruitment plot is estimated. This estimate the origin, then Fmed is likely to estimate the criti- of fishing mortality was named Fhigh by the ICES cal level of fishing mortality rather than a safe Methods Working Group, when the problem was level. Finally, if a stock suffers very high recruit- considered. Associated estimates Fmed and Flow ment variability, the estimate of Fhigh will be in- were also defined, corresponding to the median flated by that variability, as the 10th percentile is slope and that exceeded by 90% of the data points. pushed away from the median, and may not be con-

It should be noted that Flow does not always exist – servative. Finally, like the other biological refer- the biomass-per recruit obtained sometimes ex- ence points, these measures depend on the vital ceeds that corresponding to zero fishing mortality. parameters of the stock, and are in fact especially

There is, of course, no guarantee that Fhigh is a sensitive to the maturity ogive. If this has changed good estimate of the slope at the origin of the with time, the estimates may be biased. These stock–recruitment curve, even assuming that measures are therefore far from perfect, and are such a quantity is well defined: the procedure is a certainly not magical solutions to the problem of practical ad hoc procedure rather than subtle sci- providing an early warning of stock collapse. They ence. Nevertheless, the rationale for using Fhigh as must therefore be used with caution, but they do at an estimate of a level of F likely to involve an ex- least attempt to address the problem in a syste- cessive risk of stock collapse is quite plausible, and matic way, and are probably the best blunt may be stated as follows: given the data available, instruments available at the present time (see the levels of recruitment-per-unit-biomass need to also Myers, Chapter 6, Volume 1, for further maintain the stock under levels of fishing mortal- discussion). ity exceeding Fhigh have been observed in only one The problem of estimating a minimum tolera- year in ten, and it would be imprudent to assume ble spawning stock size is, if anything, even more that they could be generated routinely. Converse- difficult, and until recently relied almost entirely ly, for Flow one can say that the required levels of on subjective estimates from stock–recruitment R/SSB have been generated in most years, and plots. Serebryakov (1990) has, however, proposed these should be little risk of stock collapse. Fmed a solution to this problem which is very much divides the region of high risk from that of low in the spirit of the Fhigh approach. He suggests that risk. At or below Fmed the risk should be acceptable we identify the critical stock size as the smallest – not more than 50%, but above Fmed it begins to stock at which there is a reasonable chance of increase rapidly. This line of argument was pur- getting a good recruitment. The rationale for this sued by Sissenwine and Shepherd (1987), where it is that it invariably requires one or more good was argued that Fmed (there called Frep) could serve year classes to initiate a stock recovery. This stock as a target for management, whilst Fhigh is clearly size may be estimated from the intersection of an indicator of danger, to be avoided, and certainly the upper 10th percentile of the slope R/SSB, as not a target. used for the estimation of Fhigh, and the upper 10th Several authors have pointed out that the esti- percentile of the recruitment values, represent- mates of Fhigh and Fmed that one obtains depend to ing a typical good year class. At this stock size, Dynamic Pool Models II: Forecasting 183 good survival (high R/SSB) can still generate a mean rather than the median level, and this sup- good year class (high R). At lower stock sizes it plies a purely stochastic mechanism for regula- cannot, with reasonable probability (of the order of tion, which relies entirely upon the variability for 1 in 10). its existence. The potential efficacy of the mecha- This provides a sensible rationale for estimat- nism has been confirmed by simulation. This is, ing a critical stock size in a reasonably robust way. in many respects, a rather plausible description It has been used by the ICES North Sea Roundfish of the recruitment process (see Shepherd and Working Group (Anon. 1991), and for the North Cushing 1990 for a full discussion). At present it Sea cod stock it gave the same result as previous can only be regarded as hypothetical, but it may purely subjective evaluations. Whether or not this perhaps serve as a warning that strong assump- constitutes support for the idea is arguable! The tions about the nature of existence of stock– method has not yet been extensively field-tested, recruitment relationships are unwarranted, and nor have the consequences of using it for manage- that the whole subject needs to be viewed with ment been thoroughly simulated. At the least, it healthy scepticism. provides a sensible and objective way of estimating some low-ish value of stock size, within the range 8.2.6 Discussion observed, below which it may be imprudent to go. This is better than nothing. The abundance of fish stocks is driven to two main factors: the recruitment of young fish, and mortal- 8.2.5 Stochastic stock–recruitment ity, which is usually dominated by the effects of relationships fishing. If the recruitment is taken as given, it is not too difficult to assess the effects of varying the All the discussion in this chapter so far has implic- intensity and nature of fishing. Such yield-per- itly assumed that there is some compensatory rela- recruit calculations have been for many years the tionship between recruitment and spawning stock main tool for determining the appropriateness of size, even though this may be comprehensively various amounts and types of fishing, and remain obscured by the high variability of recruitment useful for that purpose today. (Myers, Chapter 6, Volume 1). Even this assump- However, recruitment to fish stocks is far from tion may, however, be questionable. The possi- constant in practice, but varies on all time-scales, bility that there is no such deterministic com- from years through to centuries. The reasons for pensatory relationship at all has been explored these variations remain to a large extent unknown. by Shepherd and Cushing (1990). The somewhat This severely compromises the usefulness of long- subversive idea is that one might have only vari- term forecasts for fish stocks, which must invari- ability around a constant median level of R/SSB. ably be heavily qualified with ‘other things being On the face of it, this would not supply a regulatory equal’. In particular, in most cases there is no clear mechanism. However, if the frequency distribu- indication of the effects of spawning stock size on tion of the variability is sufficiently skewed, and recruitment, and this is particularly important if the variability increases as the stock size de- because it is intimately connected with the possi- creases, as has often been suggested, then the arith- bility of stock collapse under fishing. Further- metic mean recruitment would be inflated above more, the interactions between species make the the median level at low stock sizes, and would thus absolute size of recruitment important – and may follow a compensatory relationship, of an unusual make the ‘other things’ unequal after all. This form, even though the median does not. The stock issue is discussed further in Section 8.3 below. size, at fixed fishing mortality, is of course just Given all these problems, it is clear that long- some weighted running arithmetic mean of term forecasting is a difficult area, which must be recruitment, and so responds like the arithmetic approached with considerable caution and mature 184 Chapter 8 judgement. The attempt is however essential, predictions. This was seen particularly strongly in since it is only in the long-term that the effects of the late 1980s when the northeast Arctic cod failed management are fully expressed, and any attempt to grow as expected (ICES 1990). However, in gene- to manage a fishery without the perspective of the ral the use of multispecies considerations is the long-term effects would be foolhardy. The assess- exception rather than the rule for short-term ment of the long-term effects is also the arena in predictions. which the skill and judgement of the fisheries In practice the same could also be said for long- scientist is most crucial. The short-term effects of term predictions. However, it is far less certain most fishery management actions are fairly obvi- that ignoring multispecies effects is sensible for ous to all concerned. The long-term effects are medium- to long-term predictions. Predation mor- however much less obvious, more subtle, and tality can make considerable changes to average often counter-intuitive. The assessment, and levels of natural mortality rate on pre-recruit ages explanation of these effects is arguably the most of larger fish such as the North Sea cod depicted in important, and probably the most difficult task Fig. 7.14, and it affects all ages of smaller species. which fisheries scientists are called upon to It is well known that increasing natural mortality undertake. alters the shape of yield per recruit curves. It will also affect the number of recruits living long enough to form the spawning biomass. 8.3 MULTISPECIES The equations for multispecies cohort analysis FORECASTS and the equivalent equations for multispecies VPA can be reversed and long-term simulations It was noted in Chapter 7, this volume, that, while produced, to predict how the equilibrium yield in principle changes in predation mortality might will change under various exploitation regimes for be considered in some short-term predictions, in the fish stocks included. Since fishing mortality practice for the larger species rather little is to be rate might alter, or exploitation patterns may be gained from doing so. The reason is that for these changed on any of the predator or prey stocks, a species predation mortality acts mostly on pre- wide range of scenarios is possible and the compu- recruit ages and their numbers can often be esti- tations are quite complex. A computer model, mated from research surveys at a time after the MSFOR, which makes long-term predictions major part of the predation mortality has occurred. based upon the equations and estimates of In areas with important fisheries for prey species MSVPA, was written by Sparre. It was tested and the need for considering the effect of predation used extensively by the ICES Multispecies Assess- mortality rate on short-term yield is likely to be ment Working Group (see ICES 1986). greater. As an example of this, routine calculations The wide range of possible scenarios possible of predation losses are made in the case of the with a multispecies prediction on a number of northeast Arctic capelin TAC (ICES 2000). stocks makes it rather difficult to provide under- Multispecies effects may, however, affect the standing of how the system may behave. However, growth of predator species and also their reproduc- we may generalize to a certain extent. Yield-per- tive capability. For example, Marshall et al. (1998) recruit curves typically become flatter-topped and showed that condition and fecundity in northeast their maximum moves to higher values of fishing Arctic cod was determined to a large extent by mortality rate as levels of natural mortality in- capelin abundance. Effects on growth are particu- crease on recruited ages. When natural mortality larly likely in areas heavily dependent on one key increases on pre-recruit ages the effect is to reduce prey species such as capelin in the northeast the numbers of recruits entering the fishery. This Arctic, Iceland and Newfoundland. Weight-at-age would generally depress yield-per-recruit curves if is a significant component of catch prediction so the constant recruitment of the model was at a changes in growth are likely to affect short-term younger age than some of those that the predation Dynamic Pool Models II: Forecasting 185 mortality acted on. In either case yield would seem ences are particularly large between the two runs likely to be lower when predation mortality is for haddock and herring. This largely resulted from high. This is likely to have occurred in periods apparent changes in predation patterns of saithe when fishing mortality was low. Thus the pre- between the two stomach-sampling data sets. dictions of single-species yield-per-recruit curves A further concern with such predictions is that are likely to be unsound when fishing mortality (like single-species yield-per-recruit) they do not rate is typically low throughout an ecosystem. necessarily include stock–recruitment considera- We may therefore conclude that predation mortal- tions. However, since multispecies models gener- ity generated from multispecies interactions leads ally predict asymptotic yield curves because of the to flatter-topped yield-per-recruit curves with buffering effect of predation, it is clearly even more maxima at higher levels of fishing mortality. important that stock–recruitment effects are in- This is illustrated by Fig. 8.10. This shows the cluded in multispecies models. If these are includ- calculated change in yield per average recruit for ed then it seems likely that yield curves would North Sea fish stocks that would occur if fishing initially increase more slowly with increasing mortality rate was lowered by 10% on all species. fishing mortality than a single-species yield curve The three lines show the changes expected when due to predation mortality effects. However, once predation mortality rates were based upon the recruitment overfishing occurred, the decrease in 1981 stomach-sampling data, the 1991 stomach- yield would be far more rapid than a single-species sampling data and a key run based upon both sets yield-per-recruit curve would predict. of stomach-sampling data. This illustrates that This pattern is illustrated in Fig. 8.11, which for under a multispecies model decreasing fishing the northeast Arctic cod shows the 7-year running mortality decreases yield for most of these stocks. average of yield plotted on the 7-year running This is in sharp contrast with current long-term average of fishing mortality rate, transformed to single-species predictions for these stocks which the yield/biomass ratio, expressed as the harvest would predict most yields increasing with de- percentage. This is contrasted with a single- creased fishing intensity. The figure also indicates species yield-per-recruit curve for the same stock, that the predictions are affected by the stomach scaled to the same maximum yield. It is apparent data set used in MSVPA and the consequent choice that the slope of actual yield at low fishing mortal- of suitability with which to run MSFOR. Differ- ity, as indicated by the running average curve, is

1000 5 900 800 1991 700 0 1981 600 Key run 500 7-year average catch 400 1930–6 –5 300 Scaled yield-per-

Yield and scaled Y/R recruit curve 200 100 1986–91 After Pope 1979 –10 0 0 102030405060708090 Percentage caught –15 Cod Whiting Haddock Herring Sprat N. Pout Sandeel Fig. 8.11 Yield-per-recruit curve for the northeast Arctic cod compared to a 7-year running average of Fig. 8.10 Yield of North Sea species after a general yield-against-harvest rate, and to an early production reduction of 10% in all fishing mortality. N = Norway. curve fit due to Pope (1979). 186 Chapter 8 shallower than the response predicted with a sin- of the compensatory effects that multispecies gle-species yield-per-recruit curve. Predation mor- models and stock–recruitment models attempt to tality (which is mostly caused by cannibalism in provide. However, the formulation of these sim- this case), and density-dependent growth seem the pler models is helped by more detailed understand- likely cause of this discrepancy. It is also apparent ing of the biology. Moreover, they are undoubtedly that the actual yield has dropped away very precipi- best fitted to the solid estimates of past time-series tately at higher harvest rates and this is doubtless of mortality and biomass provided by age-based es- due to stock–recruitment effects, which might timation techniques such as VPA, rather than to have been exacerbated by reductions in prey avail- the less reliable estimates derived directly from ability. The figure also shows a surplus production CPUE data. curve (Pope 1979) fitted to the biomass and harvest rates estimates of VPA. While not exact, this makes more sensible management predictions 8.4 CONCLUSIONS than the single-species yield-per-recruit curve and it is notable that it did so before the steep decline in Short-term catch forecasts are required for choos- yield occurred in the 1980s. ing among the options for managing stocks in Because of the likely changes in growth and the near future, in pursuit of whatever long-term stock-recruitment effects on yield functions, the strategy has been adopted. They are reasonably ICES Multispecies Working Group has usually straightforward calculations, which require care declined to predict long-term yield which takes and attention to detail, but few large assumptions the North Sea stocks far from their average situa- about aspects which are not well known and tion. They have thus usually just indicated the reasonably well understood. Moreover, they are changes that might result from changing effort by not much affected by multispecies biological 10% on seven broadly defined fishing fleets. This interactions, although they do depend on the tech- produces what may be converted to a Jacobian nical interactions caused by several species being matrix ∂Y(s,f)/∂E(g) of yield Y(s,f) of species (s) by caught together in mixed fisheries. fleet (f) relative to a change of effort E(g) in another Conversely, the long-term forecasts of catch fleet (g). This provides information on the local and biomass, which are needed to choose among slope of the multispecies yield per average recruit possible management strategies, are sensitive to a surface. Interestingly this and contemporary yield number of factors that are rarely well known provides just sufficient information to fit a multi- or adequately described. These include systematic species production model which can be used to variations of growth and maturity rates with stock predict yield in the near vicinity and can also be size and environmental conditions, variations used to predict the position of multispecies of natural mortality rates as a result of biological analogues to the more usual reference points (ICES interactions, especially predation, and of course 1989). the dependence of recruitment on both stock size In conclusion, multispecies effects sometimes and environmental factors, which are for all practical affect short-term predictions and can undoubtedly purposes unpredictable. For these reasons all such have a strong effect on long-term predictions. long-term calculations should not be regarded as However, the increasing detail of multispecies actual predictions of the future state of the stocks models can mean that while they may improve on concerned, except in a very general sense. So far as single-species models, particularly in the low fish- possible they should be presented as comparisons ing mortality range, their long-term predictions of the effects of following alternative management may be very variable. There may be an argument options, on the assumption that various other for making long-term predictions using simple uncertain factors remain equal, and suitable production models which implicitly include many Health Warnings should be attached. Even this is Dynamic Pool Models II: Forecasting 187 not enough, because some factors, such as preda- parent stock. Journal of the Fisheries Research Board tion rates, depend on the absolute sizes of future of Canada 30, 1965–76. stocks, which themselves also depend on uncertain Gulland, J.A. (1983) Fish Stock Assessment. Chichester, UK: Wiley. stock–recruitment relationships. Management de- Hilborn, R. and Walters, C.J. (1992) Quantitative Fish- cisions can only be based on our best estimates of eries Stock Assessment: Choice, Dynamics and Un- the future, but because these are inevitably uncer- certainty. New York: Chapman & Hall. tain, they should therefore also be informed by the ICES (1986) Report of the ad hoc Multispecies Assess- precautionary principle. It is ironical that in the ment Working Group. ICES CM, 1987/Assess: 9. field of fisheries management, where this principle ICES (1989) Report of the Multispecies Assessment is highly appropriate because we know from past Working Group. ICES CM, 1989/Assess: 20. ICES (1990) Report of the Arctic Fisheries Working experience that fish stocks can and do collapse, the Group. ICES CM, 1990/Assess: 4. default option has usually been to allow continued ICES (2000) Report of the Northern Pelagic and Blue or increased exploitation until there is incontro- Whiting Fisheries Working Group. ICES CM, vertible evidence of a problem: just the opposite of 2000/ACFM: 16. what is needed. A preference for policies that will Marshall, C.T., Kjesbu, O.S., Yaragina, N.A., Solemdal, P. be effective in maintaining or increasing spawning and Ulltang, Ø. (1998) Is spawner biomass a sensitive measure of the reproductive and recruitment potential stock sizes should therefore be incorporated in the of Northeast Arctic cod? Canadian Journal of Fisheries decision-making process. and Aquatic Sciences 55, 1766–83. However, because of the possible imprecision Pope, J.G. (1979) Population dynamics and management: of short-term forecasts, and the major uncertain- current status and future trends. Investigación ties associated with long-term forecasts, it is desir- Pesquera 43 (1), 201–21. able that management targets should also be both Quinn, T.J. and Deriso, R.B. (1999) Quantitative Fish progressive, involving moderate changes each year Dynamics. Oxford: Oxford University Press. Ricker, W.E. (1975) Computation and interpretation of in what has been assessed to be the right direction, biological statistics of fish populations. Bulletin of the and adaptive, so that the long-term objectives can Fisheries Research Board of Canada 191, 382. be updated continually in the light of new informa- Schnute, J.T. (1985) A general theory for the analysis of tion, without causing major upheavals, except in catch and effort data. Canadian Journal of Fisheries crisis situations when stock collapse is considered and Aquatic Sciences 42 (3), 414–29. to be imminent. Management strategies can be de- Serebryakov, V.P. (1990) Population fecundity and repro- signed to promote early action to avoid difficulties, ductive capacity of some food fishes in relation to year- class-strength fluctuations. J. Cons. Int. Explor. Mer. and these should be preferred, because it is ex- 47, 267–72. tremely difficult to assist stocks by anything other Shepherd, J.G. (1982) A versatile new stock and recruit- than draconian measures, once they have been ment relationship for fisheries, and the construction of severely depleted. sustainable yield curves. J. Cons. Int. Explor. Mer. 40 (1), 67–75. Shepherd, J.G. (1997) Prediction of year-class strength by calibration regression analysis of multiple recruit REFERENCES index series. ICES Journal of Marine Science 54, 741–52. Anon. (1991) Report of the North Sea Roundfish Working Shepherd, J.G. and Cushing, D.H. (1990) Regulation in Group. ICES CM, 1991/Assess: 4. fish populations: myth or mirage? Philosophical Beverton, R.J.H. and Holt, S.J. (1957) On the dynamics of Transactions of the Royal Society, London, B330, exploited fish populations. Fishery investigations. 151–64. London: HMSO, Ser. 2 (19). Shepherd, J.G. and Nicholson, M.D. (1986) Use and abuse Cook, R.M., Sinclair, A. and Stefansson, G. (1997) Poten- of multiplicative models in the analysis of catch-at-age tial collapse of North Sea cod stocks. Nature 385 data. The Statistician 35, 221–7. (6616), 521–2. Sissenwine, M.P. and Shepherd, J.G. (1987) An alterna- Cushing, D.H. (1973) The dependence of recruitment on tive perspective on recruitment overfishing and bio- 188 Chapter 8

logical reference points. Canadian Journal of Fisheries Thompson, W.F. and Bell, H. (1934) Biological statistics and Aquatic Sciences 44, 913–18. of the Pacific halibut fishery. Part 2: Effect of changes Sun, M. (1989) Simulation testing of the performance in intensity on total yield, and yield per unit gear. of methods for fish stock assessment and short- Report of the International Fisheries Commission term forecasts. PhD thesis, University of East Anglia. (No. 8). 9 A Bumpy Old Road: Size-based Methods in Fisheries Assessment

TONY J. PITCHER

9.1 INTRODUCTION and so, in one sense, it matters little how they are obtained provided they are accurate (Rosenberg If we were to take a thousand humans visiting and Beddington 1988). When properly used, size- Rome at random and line them up by rows of based methods should lead to the same estimates approximately equal height in St Peter’s Square, of these parameters as other techniques, although we could attempt a size–frequency analysis simi- the sources and impacts of uncertainty are dif- lar to that used in fish populations. The aim is to ferent. In some cases, as shown below, size-based estimate mortality and growth from the relative methods have advantages over, or can comple- frequency in the size classes. But this short-cut ment, conventional estimates based on direct human demographic analysis would fail for two ageing. reasons. First, humans stop growing in height in In fish, length is easier to measure accurately their late teens, and subsequently live for another than weight (or strictly, mass), so the term ‘length- 40 years or so at the same height, except for a slight based methods’ is in general use. Lengths tend to shrinkage when very old. Secondly, humans breed increase smoothly throughout a fish’s lifespan, ex- continuously and are therefore not born in discrete cept in very old fish that are near their asymptotic cohorts. In fact, present size–frequency analyses size (L• or W• – see Jobling, Chapter 5, Volume 1). would fail in almost all mammals for one or other Huge old cod (Gadus morhua) like this greeted of these reasons. But fish are generally born in dis- Cabot when he discovered ‘New Founde Lande’ in crete, often annual, cohorts, following an annual 1497, but they are hardly ever seen in any fisheries or seasonal breeding season, and individuals grow today. In fact, weights are an even better guide to in size throughout life towards an asymptotic size. age than lengths, but are rarely used in analyses, The growth of fish is usually well approximated by mainly because they are harder and slower to mea- a von Bertalanffy curve (see Jobling, Chapter 5, Vol- sure accurately. ume 1). Size–frequency analysis looks for peaks of Length–frequency plots are bumpy old things numbers in the size classes to estimate the mean because they reflect actual variance in size with sizes of successive cohorts at integer intervals of ages as a consequence of individual differences in age, and at the relative numbers in these cohorts to growth rate, success in food acquisition and assi- estimate total mortality rates. milation, a range of individual birth dates in one, Size-based methods can be used to obtain esti- or sometimes more, cohorts, and inaccuracies in mates of growth and mortality when fish are diffi- measurement. As we will see, the net result of all cult, or too expensive, to age using hard parts. these uncertainties makes not only for bumpy old Growth and mortality themselves are employed in plots, but also means that approximate, ‘quick and assessments of the exploitation status of fisheries, dirty’ or ad hoc, length–frequency analyses tend 190 Chapter 9 still to be useful checks. Analytical methods for The average sizes and relative abundance of the length–frequency data have a long pedigree in fish- component cohorts provide measures of growth eries science, but none of them work as well as one and mortality. First, if we can follow the mean would wish, and as a consequence many fisheries sizes in a series of samples, we can estimate researchers have dabbled in the sport of inventing growth. Secondly, if we can follow the changes in new methods at some stage or another of their numbers of a cohort with time, provided that the careers. changes accurately mirror changes in the underly- In fact, length-based analysis still presents a ing population, we can estimate mortality rates. bumpy old road for anyone wishing to employ Variation among individuals in mortality and these methods. Hence, this chapter reviews the growth can be thought of as ‘smearing’ the original underlying principles behind the main methods cohort structure. that have stood the test of time, and especially those that are amenable to the revolution that has Distributions quietly occurred over the past decade – the simple use of spreadsheets in the analysis of fisheries data. It is generally assumed that the variation in Emphasizing utility over elegance, spreadsheets length of any one cohort follows a normal distri- mean that almost any competent general fisheries bution. The expected frequency, f, at length L for scientist can make a rigorous analysis without a normal distribution of mean and standard training in the use of expensive software. deviation, is fL()ms, = [] Nw ()2 ps2 ¥-exp05 . []()L -ms2 (9.1) 9.2 AGE AND MORTALITY []{} METHODS where N = sample size, m is the mean length, s is 9.2.1 A disappearing act: modes the standard deviation of the lengths, w is the class and growth models width and L is the mid-point of the class. In fish- eries work, the assumption of normality should The sizes of fish of similar age in a cohort vary be tested using a subsample of individuals aged about a mean. Fish populations usually comprise by conventional means. A ‘quick and dirty’ test several such cohorts, which are mixed together for normality can be performed using probability in a sample. If we know the shape of the size distri- paper, or it can be tested more formally using butions of the cohorts, we can try to dissect a skew and kurtosis coefficients (or very rigorously mixed sample into its constituent cohorts. Size- using goodness-of-fit tests between the data and frequency analysis thus provides a method of the expecteds for the normal frequencies in each ageing fish without the difficulties of preparing data-length class). Other distributions are some- and reading hard parts like scales or otoliths, and times employed. The log normal, in which the without having to kill the fish sampled. lengths, means and standard deviations are trans- Variation in length of fish of a given age gener- formed to logs, may be appropriate for weight- ally follows a statistical normal distribution, frequency analysis and sometimes for length although other distributions, like the log normal frequencies. The gamma distribution is also some- or the gamma may also be employed. Length- times used. frequency plots from a sample of a fish population are therefore mixtures of a series of overlapping Parameters to estimate length distributions (Everitt and Hand 1981). Length–frequency analysis aims to dissect, to For a mixture of normal distributions we set ‘decompose’ or ‘deconvolute’, the mixture into its N to total sample size and obtain the expected components. frequency at length L as: Size-based Methods in Fisheries Assessment 191

2 may be clear and evident things to the human eye, fLii= Â{[] Nwp ()2ps 2 but with only small changes in the fish population ¥-[]exp{}05 . []()L -msii} (9.2) parameters, or to the sampling procedure, they can be surprisingly ephemeral. where N is now total sample size, pi is the propor- The conditions under which modes appear have tion of this total in the ith age group, mi is the mean been formally investigated. For two components of the ith age group, and si is the standard deviation in a mixture, Behboodian (1970) showed that sepa- of the ith age group. For h components, the prob- rate modes will be seen (bimodality) if: lem for length–frequency analysis is therefore to estimate the sets of proportions, means and stan- mm12->2min{} sm 12 , (9.3) dard deviations. The pis must sum to one, so we have (3h - 1) parameters to estimate: but even then, they will not necessarily be clear to the eye for small sample sizes. p ; s ; m ; 1 1 1 Figure 9.1(a) to (e) illustrates how modes can p ; s ; m ; 2 2 2 appear and disappear in length–frequency plots p ; s ; m ; 3 3 3 which are the overall sum of underlying normally- ...... distributed cohort components in a mixture. The p ; s ; m ; h h h values were generated from a mixture of four Statisticians have shown that mixtures of normal normal components in proportion of abundance distributions are ‘identifiable’ (Yakowitz 1969): 4:3:2:1. The resulting overall length–frequency that is, we can in principle determine all the para- plot is unimodal in Fig. 9.1(a) but modes begin to meters in the mixture, provided that the assump- appear as the means get further apart in Fig 9.1(b) tion of normality is valid and provided we know and (c). Holding the means at 10 units apart, the the combined probability exactly. In practice, of modes vanish again when the standard deviations course, we only have the data histogram to esti- are increased in Fig 9.1(d) and (e). mate the latter. There is a more detailed discussion Following these rules, three main factors in the of this point in Macdonald and Pitcher (1979). fish population conspire to reduce the separability of modes in length–frequency data. First, if fish grow according to the von The appearance of modes Bertalanffy curve, as they approach L•, the Cohorts of fish which recruit at different times are cohort means get closer together and are there- consequently separated in mean size. The appear- fore less likely to reveal modes. This is known ance of separate modes in a size–frequency plot of a as the ‘pile-up’ effect. The ‘pile-up’ effect is sample taken from the whole population has long illustrated in Fig. 9.2 and may be investigated been interpreted by ecologists as revealing age using the animated spreadsheet available at groups (e.g. Petersen 1891). Within any one cohort there will be a spread of sizes resulting from differ- Secondly, variance in length-at-age increases ent birth dates and individual growth rates. This with length as fish get larger and approach L•, and spread may obscure the modal sizes of the separate so age groups are more likely to overlap. For cohorts. randomly-varying L•, Rosenberg and Beddington The procedure of estimating growth and mor- (1987) show that modes will appear in a two-cohort tality from a series of samples by tracking the mixture if: modes of each cohort has been termed ‘modal pro- gression analysis’. There are a number of graphical {}Lktt• exp[]--()0 ¥-[] exp() k-1 2 2 techniques for achieving this aim, but they >---21s Lktt{}exp[]()0 (9.4) work well only when definite modes appear in the length–frequency plot. The problem is that modes where s2L is the variance at length L. This function 192 Chapter 9

(a) (b) 80 80

60 60

40 40 Frequency Frequency 20 20

0 0 Length Length (c) (d) 80 80

60 60

40 40 Frequency Frequency 20 20

0 0 Length Length (e) 80

60

40 Frequency 20

0 Length

Fig. 9.1 Diagrams show the normally distributed cohort components (thin line) in length–frequency mixtures (thick line) generated from an algorithm (as described in the text). (a) The means of the four normally distributed components are 5 units apart and the standard deviations are set to 2.5. The overall envelope, representing the size–frequency plot, is unimodal. The overlap index V = 0.49. (b) The means of the four normally distributed compo- nents are increased to 7 units apart with the standard deviations remaining at 2.5. Separate modes are beginning to appear in the overall envelope. The overlap index V = 0.29. (c) The means of the four normally distributed components are further increased to 10 units apart with the standard deviations remaining at 2.5. Separate modes are clearly evident in the overall envelope. The overlap index V = 0.02. (d) The means of the four normally distributed components remain 10 units apart but the standard deviations are increased to 3.5. The modes, especially for the less abundant older age groups, are less evident in the overall envelope. The overlap index V = 0.27. (e) The means of the four normally distributed components remain at 10 units apart while the standard deviations are now further increased to 4.5. Separate modes have disappeared the overlap index V = 0.43. Size-based Methods in Fisheries Assessment 193

L inf Length

Fig. 9.2 Diagram illustrating the ‘pile-up’ effect. The larger diagram shows normal distributions of 0510 length around the mean lengths Frequency Age of 10 successive annual cohorts. Inset: length-at-age projected (horizontally) from each mean length at integer age (circles) on a von Bertalanffy growth curve. Source: diagram taken from an animated spreadsheet available from www.fisheries.ubc.ca/ projects/lbased.htm. Length will vary with length, so that separate modes are extended season in some tropical fisheries for less likely at greater lengths if s2L increases. which length-based methods are otherwise ideal. Unfortunately, the way in which s2L changes If modes are present, they probably reveal co- with size can be quite complex and there has been horts, but if modes are absent, there can still be no rigorous investigation using actual fish growth. several cohorts present. This means that the older Rosenberg and Beddington (1987) show that if dif- name for these methods, ‘polymodal analysis’, can ferences in L• are the main source of variation be- be misleading. If modes appear, the simple graphi- tween individual fish, s2L increases with fish size. cal or approximate computer methods will give On the other hand, if most variation between fish good results. If there are no modes, one of the sta- is in the growth parameter k, s2L peaks at about tistical methods will be needed. In their review of half L• and then drops as fish approach the asymp- length-based methods, Rosenberg and Beddington totic size. Empirical data usually show variance in (1988) recommend the greater use of formal statis- length increasing with size, so that for sizes up to tical methods, which are not so dependent upon

0.7 L• the assumption of a constant coefficient of the appearance of modes. variation of length (COV-l) seems reasonable. The simulated length–frequency data used in this An index of overlap as a guide to paper uses a COV-l which falls almost impercepti- the appearance of modes bly with increasing length at first, but drops rapid- ly as L• is approached. This section presents an index of overlap which The first two problems above affect older age will, in conjunction with a consideration of sam- groups more seriously. But the third mechanism ple size, indicate whether modes are likely to which may obscure modes can affect young and old appear. This in turn will allow the researcher age groups alike. If recruitment of cohorts is con- to decide whether simple graphical or ad hoc tinuous, or extended over a large portion of the methods are likely to be adequate. year, the variance in length by age group will be The first step in the calculation of an index, is to large and modes may be obscured from young ages decide roughly what the likely means and standard on. Unfortunately, recruitment may occur over an deviations of the proposed age groups are. The 194 Chapter 9 overlap index can then be calculated as follows, The first of these can be accommodated by most a quick task using any modern spreadsheet: length-based methods, provided that the variance is not too large. But the second is a major problem ih=-1 for all the methods, as it tends to destroy cohort Vqq=+Â {[]()msmii--() i++11 s i i=1 structure. []()msmsii+ qq--() ii} h (9.5) A further assumption is that the length- where q = 1.96 to give 95% limits, and h = number frequency data in your sample fully represent the of age groups. This is then repeated in order to com- length classes in the fish stock. If they do not, then pare several alternative hypotheses. the sample data will need to be adjusted to com- The index V reflects the average proportion pensate for the selectivity of the sample gear. A by which the 95% zone (i.e. 19 out of 20 fish of neat series of tests for this employs the relation- this age) for the age group is overlapped by the ship among length at maturity, Lm, age of maxi- 95% zone for the next age group. When V is mum yield-per-recruit, Lopt and temperature in negative, there is a very wide separation of the order to evaluate the validity of the length– age groups. When V is greater than about 0.25, frequency sample (Froese and Binholan 2000). For modes disappear. The V for individual age groups example, these authors show how a trawl survey of can also be usefully examined where the separa- Nile perch (Lates nilotica) taken in Lake Victoria tion of adjacent ages differs across the length– in 1982 missed all fish larger than 100cm. Since frequency plot. Example values of V are given in this is far less than Lopt = 136cm, the survey Fig. 9.1(a) to (e). length–frequency data could not be used for assess- ment of the perch population at that time. 9.2.2 Assumptions of The starting point in all analyses is the length–frequency analysis length–frequency distribution, adjusted if neces- sary, with known class width and class boundaries. For length-based methods to work, fish must It is worthwhile taking a lot of care over the recruit in discrete cohorts. There is no obvious class boundaries: lower bound, mid-point and way that length-based methods could provide both upper bound of the classes are all used in different growth and mortality estimates for continuously methods. recruiting populations, like the human example given earlier. For a caveat see the Conclusions 9.2.3 Classification of section of this chapter. Discrete cohorts usually length–frequency analysis methods derive from separate spawning seasons, but there may be more than one of these per year. The co- Methods may be divided into parametric and horts must remain discrete as the fish grow older. non-parametric. Another classification is into This requirement may be relaxed to a certain simple ad hoc methods, which are often essen- extent for different methods of analysis, but, gen- tially graphical or non-parametric, and rigorous erally, the methods work better the more discrete statistical estimation methods, which are usually the cohorts and the more separate they remain as parametric. they get older. This implies that the growth of indi- Parametric methods depend upon estimating viduals in a cohort should be similar, i.e. that vari- the means, standard deviations and proportions or ability in growth rates among individuals of the numbers in each of the cohorts in the mixed sam- same age is not large. ple. These are the parameters of the size-frequency So there are two sources of variation in size distributions, hence the term parametric. The size within a cohort of fish: distributions are generally taken as normal, but log • different birth dates within the spawning normal and gamma distributions may also be em- season; ployed (Macdonald and Pitcher 1979). The meth- • different growth rates among individuals. ods include both graphical (e.g. probability plots) Size-based Methods in Fisheries Assessment 195 and computational (e.g. mixture analysis) meth- from a series of formal single-sample estimations: ods, but all make strong assumptions about distri- a clear example is discussed by Sparre and Venema butions. In parametric methods, the number of (1992). cohorts generally has to be determined by the user, Gulland and Rosenberg (1990) outlined simple and several scenarios may have to be compared. interpretations that may be made from visual in- Non-parametric methods do not depend upon spection of length–frequency plots. Type A, a estimating the parameters of the cohort distribu- single mode that stays in the same place through tions directly. So they make only weak assump- time, can be produced by gear with high selecti- tions about the distribution of sizes within the vity, such as gill-nets, or by fish, for example cohorts, that they are roughly distributed about yellowfin tuna (Thunnus albacares), that migrate some modal or central value, and hence are analo- with age. The authors say that not much can be gous to non-parametric statistics. The modal done with type A although see the conclusion sec- lengths of each cohort are fixed to lie upon a curve tion of this chapter. Type B, a single mode moving described by a growth model. Generally the von steadily upwards, is typical of single-cohort fish- Bertalanffy model is used, but other models such eries such as prawns or squid, which are good can- as a seasonal growth model, can be employed. didates for any simple analysis. Type C, with many Hence the non-parametric methods make strong clear modes like Fig. 9.1(b) or (c), may also be a good assumptions about growth. In non-parametric subject for the classical techniques described methods, the number of cohorts is implicit in below. Type D, with smeared modes like Fig. 9.1(a) the estimates of growth model parameters, and or (e), may be hard to analyse. may be revealed when cohorts are sliced into The use of probability plots was reinvented age groups. several times by fishery workers (Harding 1949; The following sections briefly describe graphi- Cassie 1954; Harris 1968). Originally they were cal methods, most of which are parametric, three done on special probability paper, but today it is non-parametric methods, and two fully statistical easy to set them up on a spreadsheet using the parametric distribution mixture methods. Most of built-in normal distribution function. A series of them can be carried out using spreadsheets. progressively more sophisticated graphical meth- ods were based on the change of slope of a parabolic 9.2.4 Graphical methods function of frequency and length (Buchanan- Wollaston and Hodgeson 1929; Hald 1952; Tanaka A number of graphical methods aim to estimate 1962; Bhattacharya 1967). Some versions of the the age-group parameters. Graphical methods FiSAT computer package include dissection using have the advantage of being quickly performed a modification of Bhattacharya’s method (Pauly with a simple spreadsheet, or even pencil and and Caddy 1985). Taylor (1965) invented an intri- paper, and bypass statistical difficulties. In most cate method. After smoothing the data histogram, methods of this type, successive components are components are sequentially extracted from the extracted sequentially from the data. shape of the left flank of the distribution. Modal Progression Analysis, the simplest and There has been a long search for an improve- oldest method, entails the graphical joining of co- ment upon the subjectivity of the graphical meth- horts which appear as clear modes. If this can be ods, which nonetheless retain their attraction in done, it may not be worthwhile to go any further bypassing the problem of statistically estimating with more elaborate methods! Problems arise the standard deviations and proportions. In my when deciding which cohorts to join up with opinion, none of the later graphical methods offer which others. For species that actually shrink, much improvement on probability plots in decid- such as octopus, lamprey (Lampetra spp.), this ing among various alternative interpretations in may be one of the few methods applicable. Modal difficult cases, and are hence not worth the consid- progression analysis can also be used on the results erable extra effort in calculation or computing. 196 Chapter 9

9.2.5 Non-parametric methods plained sum of peaks’. This is based on how often a von Bertalanffy growth curve hits modes in the Most non-parametric methods work by scanning data. During fitting, growth curves with different a range of L• and k values and working out a parameters are run and mapped. The maximum goodness-of-fit (GOF) for each combination. The of the scoring function is chosen as the best fit. A best GOF is searched for by the user or by auto- seasonally modified growth curve can easily be matic search, or a combination of both. The best fit substituted for the standard von Bertalanffy, and, gives the estimate of growth. Mortality is esti- in fact, the same goodness-of-fit technique could mated from the numbers in the cohorts sliced be used for any growth curve or pattern. from the original size distribution using these The number of peaks is identified by those growth parameters. standing out above a 5-point moving average: Usually, these methods attempt to do this by troughs are the areas below the moving average. fitting a growth curve through a whole set of sam- The number of peaks gives the maximum ‘avail- ples taken through time. The von Bertalanffy able sum of peaks’ (ASP). A von Bertalanffy curve growth curve, or its seasonal modification, is for the specified L• and k is traced through the data ordinarily employed, although it is possible to use starting at the base of the first peak. A point is other growth models, or even empirical growth scored each time the curve hits one of the peaks: a values, but these options have rarely been used. point is deducted each time it hits a trough. This is A goodness-of-fit function based on how well the repeated for starting times equal to the base of each growth curve passes through the ‘peaks and peak. The maximum value is the ‘explained sum troughs’, is maximized for a range of values of L• of peaks’ (ESP). The goodness-of-fit function is the and k. So growth, and sometimes mortality, is ratio ESP/ASP. This process is repeated for all estimated along with the dissection of the required combinations of L• and k, the goodness- length–frequency curves. of-fit mapped and the maximum value chosen as Three of these non-parametric methods are out- giving the best growth parameters. For each com- lined in this paper: ELEFAN, originally devised by bination of k and L•, it is possible to search for the Daniel Pauly; SLCA, originated by John Shepherd; value of t0, the starting point within a year for the and the Projection Matrix method, which was growth curve, relative to the data, which maxi- developed by Andrew Rosenberg and Marinelle mizes the ESP/ASP ratio. Note that absolute ages

Basson (Rosenberg et al. 1986) from an original are needed to find the true t0. method by Shepherd (1987b). Simulations show that ELEFAN can give clear and correct answers where peaks are well separ- ated in the data, but that it tends to underestimate ELEFAN (electronic length–frequency analysis) k (Rosenberg and Beddington 1987). There are two Daniel Pauly was the first to realize the potential problems with the ELEFAN technique. First, it of this type of method, and working versions of his seems sensitive to the appearance of discrete original ELEFAN first appeared in the late 1970s modes in the data. Second, because it is an ad hoc (Pauly and David 1980). Nowadays, the modern method, it lacks explicit statistical error structure version of this length–frequency method is the and therefore provides neither standard errors of ELEFAN 1 module of the widely-used FiSAT pack- the estimates nor a guide to performance in any age distributed by FAO (Food and Agriculture situation. But the latter problem could today be Organization) (Gayanilo et al. 1996). Pauly (1987) investigated using Monte Carlo simulation has written a very clear review of the basis of methods (Hilborn and Mangel 1997). the method. Like the other non-parametric methods, How ELEFAN works. ELEFAN works by at- ELEFAN does not directly evaluate multiple tempting to find a maximum for a goodness-of-fit recruitments during a year, such as occurs in many function based on peaks and troughs: the ‘ex- tropical fisheries, although there is a recruitment Size-based Methods in Fisheries Assessment 197

pattern routine which helps to detect multiple re- (Note: this covers the likely range of t0 values). The cruitment pulses. Alternative growth models can maximum score, Sm, for the current combination be used instead of the von Bertalanffy, and one that of growth parameters is then given by: has been frequently employed is a seasonal version 22 of this growth curve. SSaSbm =+(). (9.9)

For any one pair of values of L and k, t can be SLCA (Shepherd’s • 0 easily found as: length-composition analysis)

Shepherd (1987a) introduced an objective tSbSa0 = arctan()2p . (9.10) goodness-of-fit function for detecting peaks and troughs, using a damped sine-wave function The above is repeated for all the L• and k combi- borrowed from time-series analysis of diffrac- nations under consideration, and the values of tion patterns. The damped sine-wave function goodness-of-fit, Sm, are entered into a table of emulates the decreasing spacing of mean lengths- results so that the maximum may be identified. at-age of the von Bertalanffy curve. The SLCA Contours of Sm may be mapped to avoid picking method is conceptually very similar to ELEFAN, local maxima. As with ELEFAN, the upper limit the value of a scoring function being mapped of length classes needs truncating to avoid bias against a range of values of L• and k. from the ‘pile-up’ effect. The ‘pile-up’ effect can be How SLCA works. Values of L• and k are minimized by the use of the Pauly and Arreguin- chosen for a von Bertalanffy curve. For each, length Sanchez modification. interval L, tmax and tmin are calculated as the ages Provided ages as defined by the start point in the corresponding to the start and mid-point of the analysis are known, Shepherd’s method can pro- interval using the growth equation: t is the average vide a direct estimate of t0. Published simulations of tmax and tmin. The test function TL is estimated suggest that it is more robust than the ELEFAN as: algorithm (Basson et al. 1988). Provided that the modes for the younger fish are reasonably clear in

TQQttLs= []sin()()pp¥-{} cos[]2 p() , (9.6) the samples, it seems less sensitive to the appear- ance of modes overall. But Terceiro and Idoine where Q = (tmax - tmin) and ts is the proportion of (1990) showed that SLCA suffers from the same the year since recruitment when the sample was general problems as the other methods. taken. The GOF function is then calculated over The algorithm of SLCA is firmly linked to the all length groups as: von Bertalanffy model, and it would be hard to modify it for seasonal growth or alternative growth STNt= Â[]()LLD L (9.7) models. where N is the number in each length class, and L Projection matrix method DtL is the time needed to grow through each length class: The basis of the projection matrix method (Rosenberg et al. 1986) is elegant, simple and differ-

DtLud=-1ln k ◊[]() LLLL•• - ()- (9.8) ent from the previous two methods. The numbers in any size group at time (t + 1) can be predicted where Lu is the upper bound of the length class and from the numbers in that group at time t, using Ld is the lower bound. This modification was intro- growth, mortality and recruitment from length duced by Pauly and Arreguin-Sanchez (1995). groups smaller in size:

To estimate t0, this is run with t0 set to zero, to give Sa, then again with t0 set to 0.25, giving Sb. fgZ[],,() fii tÆ fgZ{} ,,() f t+ 1 . (9.11) 198 Chapter 9

Now, if constant mortality is assumed over the any growth model is easily incorporated in the time interval, or the pattern with size is known, projection. only the growth model parameters will affect the projected numbers. Summary: non-parametric methods How the projection matrix works. First, using the first set of parameters for the growth curve, All three non-parametric methods can easily give the expected length frequency is projected for- silly answers, and should only to be applied with wards from the first sample in the set. Secondly, care and with insight of the ecology of the fish the goodness-of-fit, G, of this expected data in each under study. Using Monte Carlo simulations, Isaac class i, proji, is compared with the actual length (1990) compared how SLCA and ELEFAN perform frequencies, obsi, using least squares: under a variety of conditions and provides some helpful guidelines. For example, SLCA seems to be 2 G =-Â[]()obsii proj . (9.12) better for slow-growing and ELEFAN better for fast-growing fish. Thirdly, the projection of expected values is re- All methods based on the von Bertalanffy curve peated using the next set of values for the growth suffer from multiple optima at harmonic combina- curve parameters. This is repeated for the whole tions of k and L• (see Kleiber and Pauly 1991), range of parameters and the G goodness-of-fit which is not surprising given that a number of values tabulated and contoured so that, as with alternative fits to length–frequency data will be SLCA, the best fit can be chosen. Note that here a reasonable statistically. A degree of subjectivity is minimum value of G is sought. As with SLCA and inevitable in interpreting the GOF response sur- ELEFAN, the upper-length classes need truncat- face: all the methods can generate multiple peaks ing to avoid bias from the ‘pile-up’ effect. along ridges of high GOF (see example below) and The projection matrix method can perform simulations show that in some cases a high peak is well under a wide range of conditions (Basson et al. not the correct answer. The recommended non- 1988). Unlike all the other non-parametric and parametric approach to length–frequency analysis graphical methods, the Projection Matrix does is to try to find solutions which are robust against not rely on the appearance of ‘peaks and troughs’ the particular method used. If all three non- (modes) in the data, an advantage it shares parametric methods indicate similar peak GOFs, with parametric distribution mixture methods. then you can have some confidence in the answers. It is also robust against increase of variance Where they differ, decisions have to be based on in length with age. An advantage is that any additional knowledge of the species in question. growth model could be used for the projection. In its basic form it suffers from the same multiple 1 The far-horizon (‘rubber tuna’) problem The recruitment problem as ELEFAN and SLCA, but, far-horizon problem occurs when a good statistical like ELEFAN, it could be easily modified to deal fit is obtained at high L• and k. This fit implies very with this. rapid growth and, by analogy, a k of 2 might corre- In my opinion, of the three non-parametric spond to the ‘growth rate’ of a tuna-shaped rubber length–frequency analysis methods, the projec- balloon inflated rapidly with a tyre pump. This tion matrix, as the least ad hoc, is the one that would be an absurdly fast growth rate. A good fit would stand more statistical development. Unre- like this would imply that all the bumps and wig- solved problems in the present version appear to be gles in the length–frequency data are only noise how to evaluate shifts in the date of origin of the and there is only one, or a few, cohorts present. Co- projections, and whether projections from growth hort slicing, which assigns fish to ages (see below), year 1 should be applied to samples taken in subse- and examination of the fitted growth curve against quent calendar years. Seasonality in growth needs the data histograms will check out whether the to be included, but this should not be difficult as suggested far-horizon fit is realistic. SLCA is Size-based Methods in Fisheries Assessment 199 especially prone to this problem, and, to avoid it, it metric ratios, and so plotting published values of is recommended to scale the GOF by dividing by logL• against logk can be a good guide to the accu- (L• *k). The general remedy for all methods is to racy of estimates from length–frequency analysis. beware of far-horizon solutions unless (a) you have Figure 9.3 shows an example of such an analysis good evidence that fish actually grow that fast of small pelagic zooplanktivores from some

(some do, e.g. Coryphaena), and (b) you are satisfied African lakes (Pitcher et al. 1996). Values of L• and with the implications shown by cohort slicing. k that fall outside the boundaries of the regions characteristic of each species are unlikely to be 2 The near-horizon (‘bumpy road’) problem correct and can be avoided. It is recommended to The near-horizon problem occurs when a good sta- draw up auximetric plots like this one when per- tistical fit is obtained at very low or very high L•, forming new analyses. Phi prime values can be ob- and low k, where L• is very close to Lmax. Cohort tained for many species and locations from the slicing on these growth values will reveal many FishBase website, where there is also an online age classes. A good fit here implies that every little routine for plotting the equivalent of Fig. 9.3 bump and wiggle in the length–frequency distribu- (www.fishbase.org). tion is a cohort, meaning an absurdly slow growth rate. The only remedy is to beware of solutions An example of non-parametric analysis To illus- which are very close to Lmax and to examine the im- trate the use of the three non-parametric methods, plications with cohort slicing. In general k values this section compares the fits obtained to some less than 0.05 are suspect, but of course this can generated test data. Six bi-monthly samples of mask growth rates of fish that are long-lived and around 1500 individuals from a hypothetical popu- genuinely slow growing, such as sharks, orange lation are plotted in Fig. 9.4. Fish growth is nor- roughy (Hoplostethus atlanticus), Pacific rockfish mally distributed (coefficient of variation = 0.1) (Sebastes spp.) or sturgeon (Acipenseridae).

3 General approach to multiple optima It is Limnothrissa – Tanganyika not always easy to choose one from a set of many Limnothrissa – Kivu GOF peaks, especially if you get them from dif- 0.8 Limnothrissa – Kariba Limnothrissa – Cahora Bassa ferent methods. In the last resort you may have to 0.6 Engraulicypris – Malawi retain several peaks right through to the assess- Rastrineobola – Victoria ment stage and examine the management implica- 0.4

tions of each one. Additional information can be k brought to bear upon the problem. For example,

Log 0.2 one can look at the implications for age structure of alternative peaks using cohort slicing, or otolith 0.0 1.5 1.7 1.9 2.1 2.3 2.5 or scale readings of subsamples. Log L-inf A powerful general approach is to filter the re- –0.2 sults using data from similar species which have –0.4 been analysed elsewhere. Pauly and Munro (1984) demonstrated what they termed an ‘auximetric’ Fig. 9.3 Auximetric plot of published growth parameters of three species of small pelagic cyprinids relationship, ‘Phi prime’, between L• and k that is consistent across species: and clupeids from African lakes. Points within lines indicate values for each species consistent with the Phi prime analysis. Points outside the lines are suspect logk 2logL (9.13) F¢ = + • and should only be used with caution. Circular symbols are cyprinids; square symbols are sardines. Analysis of thousands of such relationships shows (Source: data taken from Pitcher et al. 1996, where that members of a taxon have closely similar auxi- full details of sources may be found.) 200 Chapter 9

.

Fig. 9.4 Length–frequency plots of samples taken from a hypo- thetical fish population with growth and mortality parameters as described in the text.

around a von Bertalanffy curve with L• = 1000mm other hand, ELEFAN, which has an interesting- and k = 0.1. The instantaneous mortality rate is set looking rough and ridged surface that might to 0.2 and samples are taken with equal catchabili- confuse a local search procedure, exhibits a huge ty, so that sample sizes reflect abundance. trough of very unlikely values near to the correct

Goodness-of-fit surfaces for values of L• from values that might help in focusing attention on the 40 to 1500mm, and k from 0.05 to 0.3 are drawn correct area. This example reinforces the recom- in Plate 1 for each of the three non-parametric mendation to run all three of the non-parametric length–frequency analysis methods (see colour methods on any data you may wish to analyse. plate). Coloured shading runs from red (low) to pur- ple (high). The correct value is indicated by a star on 9.2.6 Parametric methods the response surface and cross-wires on the base. All three methods exhibit the banana-shaped area of al- Parametric length–frequency methods work by ternative fits very clearly, and this is intrinsic to any calculating a GOF between the sample data and a method that attempts to fit a von Bertalanffy curve. distribution mixture specified by its component For these data, the projection matrix method is parameters. The history of these methods is re- the only one to find the true value. However, the viewed by Macdonald and Pitcher (1979). GOF is peak GOF is a small bump in the middle of a wide- calculated as the difference between the sample domed plateau, and it would not have been found data and the fitted mixture of distributions. The without very fine scale intervals in the search pro- methods usually work by searching automatically cedure. Plate 1 also demonstrates one of the prob- for a maximum GOF, but the user can also inter- lems with both ELEFAN and SLCA. Both show the vene and guide the fitting process. The number of highest GOF in banana-shaped peaks that increase component cohorts is usually the choice of the at high L• and low k (the ‘bumpy road’ effect). user, guided by the GOF values of alternatives. Al- SLCA has a smoother GOF surface than ELE- ternative fits with different numbers of compo- FAN. Hence, for SLCA, automated search proce- nent age groups can be compared: dures (such as are embedded in the LFDA package) 2 are less likely to find a spurious local peak. On the Chi square= Â{}[]() obsLLL- ff. (9.14) Size-based Methods in Fisheries Assessment 201

8 number of components. A more complex but essentially similar statistical method, MULTI- FAN (Fournier et al. 1990), gets around the prob- 6 lem by using a von Bertalanffy curve to provide the number of cohorts in a similar fashion to the non- parametric methods. In fact, results from MIX and the more complex multi-sample MULTIFAN are 4 generally very similar (Wise et al. 1994; Kerstan

Frequency 1995). Experience suggests that MIX is robust for 2 single-sample analysis, although it tends to under- estimate k (Rosenberg and Beddington 1987). Where there is a series of samples it has advantages if there is any reason to suspect that growth does not fol- 0 0 5 10 15 20 25 30 35 low a von Bertalanffy curve. This can happen in Length some fish that switch to piscivory during their lifespan (LeCren 1992). The additional work in the Fig. 9.5 Example of length–frequency analysis as a multi-sample MIX technique is to join cohorts in statistical dissection of a distribution mixture – the MIX successive samples using an MPA-like method, technique. Shaded area represents histogram of length which can be both an advantage and a disadvan- frequencies from fish sample data. Thin lines are normal tage. Modifications to the MIX approach can easily distributions; thick line is overall mixture. Parameters of the component normal distributions are adjusted incorporate information about growth (e.g. Liu until difference between overall mixture and histogram et al. 1989) either as starting values for mean is at a minimum (least squares). Source: figures cohort sizes and/or as additional constraints on the from a spreadsheet that may be obtained from fitting process. Schnute and Fournier (1980) pub- www.fisheries.ubc.ca/projects/lbased.htm. lished an alternative version of this process. In the tropics, there is often more than one co- hort recruiting each year which is a consequence of This is the basis of statistical mixture analysis, monsoon-like seasonality in productivity. For ex- originally embodied in the MIX technique ample, Koranteng and Pitcher (1987) used MIX to (Macdonald and Pitcher 1979), although its statis- analyse length–frequency data for a West African tical roots go back to Peterson (1891). The essential sparid fishery, where a cohort recruited after each calculations, using f(L) from the first two equa- of the two major upwellings each year. The plot of tions in this chapter, can today be easily performed estimated means from MIX was best joined up on spreadsheet (e.g. Fig. 9.5). using a strong assumption that there were two co- The main problem in using the MIX approach is horts per year. In fact, similar results can be ob- obtaining the number of components in the mix- tained using ELEFAN if a similar assumption is ture. The approach recommended by Macdonald made (Pauly, personal communication). and Pitcher (1979) and Macdonald and Green An example of the detailed analysis of cohorts (1988) is to get the best fit for h - 1, h and h + 1 com- of a small freshwater fish is presented in Fig. 9.6 ponents where h is the guessed number of compo- (data from Pitcher 1971). Samples were taken nents, the final choice of number of age classes using back-pack electro fishing at approximately being mainly on the basis of the minimum chi- every 4 weeks and subjected to mixture analysis square. Rosenberg and Beddington (1987) show using the MIX routines. The progression of mean that MIX growth parameter estimates are quite lengths of each cohort identified from the samples robust against small mistakes in obtaining the were traced using information from seasonal 202 Chapter 9

90

80

70

60

50 +

40 Length (mm) 30 Fig. 9.6 Example of tracing 20 cohort progressions following length–frequency analysis of sam- ples of 90–300 minnows (Phoxinus 10 phoxinus) from a 0.5km reach of the Seacourt Stream, Berkshire 0 0 16 32 48 64 80 96 112 128 (UK) in 1967–9. Mean lengths Time (weeks) estimated from each cohort are represented by different symbols: lines join up most likely progress- ions due to growth. (Source: data Jul. 67 Jul. 68 Oct. 67 Oct. 68 Jan. 68 Jan. 69 Jun. 67 Jun. 68 Jun. 69 Apr. 67 Feb. 68 Apr. 68 Feb. 69 Apr. 69 Sep. 67 Sep. 68 May 67 May 68 May 69 Dec. 67 Dec. 68 Mar. 67 Mar. 69 Mar. 68 Nov. 67 Nov. 68 Aug. 67 Aug. 68 redrawn from Pitcher 1971.) growth (see Pitcher and Macdonald 1973). The pic- (Rosenberg and Beddington 1988). This is very ture was complicated by the hatching of several co- similar to using an age–length key (Gulland and horts of eggs from separate groups of spawning fish Rosenberg 1990; Sparre and Hart, Chapter 13, this in summer. Figure 9.6 shows three such young- volume). The relative numbers of fish in the fitted of-the-year and 1+ age-group cohorts traced from mixture components give the annual mortality births in 1966 and 1967, but such multiple cohorts rates directly. were not detected in 1968. This detailed analysis of But for non-parametric methods, this informa- the growth, and mortality, of sub-cohorts from dif- tion will not be available, so the overlap classes ferent hatch-groups would have been difficult with may have to be split on an ad hocbasis to mimic the the non-parametric techniques. Overall the best changing component proportions. After slicing, advice is to plot out all means against time and age the relative numbers in adjacent cohorts may be so that effects like this can be detected. used to estimate mortality rates. A better method is to track samples through time and examine mor- 9.2.7 Cohort slicing and estimation tality rates for a cohort though time. Anomalous of mortality samples will often show up when these are plotted and can be eliminated from the analysis. Cohort slicing enables attribution of the fish An example from a Hong Kong fish, Siganus in each length group to an age cohort using the canaliculatus (white-spotted rabbitfish), is shown information on mean length-at-age. For para- in Fig. 9.7 (Pitcher et al. 1998). Combining esti- metric methods, this is easy because in overlap mates for the period of the year when both 0+ and zones the proportion of fish in each length class is 1+ cohorts were present in samples, the total mor- allocated in proportion to the relative frequency of tality rate is estimated as 7.7 (note this includes each age group in the fitted distribution mixture offshore migration as discussed below). Size-based Methods in Fisheries Assessment 203

8 There are also a range of graphical length-based Siganus 7 methods that estimate mortality directly, usually based on scaling the decline in numbers along the 6 right-hand limb of a length frequency plot with the 5 time taken to grow through each length interval. An allied method (Wetherall’s) depends on esti- N 4 mating Z/k, and hence, given k, total mortality. Log 3 These methods have been thoroughly reviewed by Sparre and Venema (1992). 2 1 9.2.8 A guide to computer packages 0 0 0.2 0.4 0.6 0.8 1 Table 9.1 tabulates available computer packages Time (year) against features discussed in this chapter, such as the strength of assumptions about distributions Fig. 9.7 Total mortality rate of Hong Kong white- and growth. Table 9.1 also shows that nearly all of spotted rabbitfish following cohort slicing of length– the published methods may these days be pro- frequency samples aged using non-parametric methods grammed into a spreadsheet such as Excel, with with an adjustment for overlap (see text). Plot shows Visual Basic routines written to step across the estimated numbers of two cohorts (open circles = 0 + co- range of required L and k values and to call the hort; closed circles = 1 + cohort) over 13 samples. Points • at right-hand side of plot are used to estimate mortality built-in Excel optimization routines (‘Solver’). The rates (solid lines). An example spreadsheet may be down- routine can record the GOF at each step for later loaded from www.fisheries.ubc.ca/projects/lbased.htm. plotting and interpretation. The advantage of this

Table 9.1 Guide to computer packages for length–frequency analysis.

Method MPA Efan SLCA Proj PPap Bhatt Nmsep Mix Mfan

ASSUMPTIONS parametric/ NP NP NP NP P P P P P non-parametric TYPE graphical/ G C C C G G C C C computational DISTRIBUTIONS weak/strong W W W W S S S S S GROWTH weak/strong W/S S S S W W W W/S S SAMPLES Single/multiple M S/M S/M M S S S S M SPREADSHEET? Hard/Easy/No Easy Easy Easy Hard Easy Easy Hard Easy No PACKAGE CODE – 1,2,5 2,5 2 – 1,5 5 3 4

Notes: Key to methods: MPA = Modal Progression Analysis (MPA); Efan = ELEFAN; SLCA = Shepherd’s Length- Composition Analysis; Proj = Projection Matrix; PPap = Probability Paper; Bhatt = Bhattacharya’s method; Nmsep = Normsep; MIX; Mfan = MULTIFAN. Key to packages: (1) Compleat Elefan (ICLARM); (2) LFDA 3.0 (RRAG); (3) MIX (Ichthus); (4) MULTIFAN (Otter); (5) FiSAT (FAO). 204 Chapter 9 approach is that the user retains complete control which dissected 12 age groups from Bluefin tuna over the algorithm and the interpretations of alter- (Thunnus thynnus) samples (Fournier et al. 1990). native maxima. One disadvantage is that some The software contains some neat graphical short packages contain undocumented improvements cuts together with powerful optimization rou- and corrections to published methods that spread- tines. MULTIFAN runs in a DOS Window in Win- sheet programmers may find it hard to obtain dows 95 of NT. A powerful modern version is information about. Moreover, many existing under development, but its website implies that it packages were generally written for DOS and most will cost over US$10000. As to costs: LFDA is free- have not been updated to Windows versions. ware, FiSAT costs US$40, MIX costs around A DOS computer package called LFDA (Length US$140, while MULTIFAN is very expensive at Frequency Data Analysis) is available from the about US$1500. Renewable Resources Assessment Group at Imperial College, London, UK (www.ic.ac.uk). It includes SLCA, the Projection Matrix and the 9.3 WHEN LENGTH IS KING ELEFAN algorithm, along with various options for estimating mortality, and has some elegant con- Some analyses may only be performed using tour plots of the GOF surfaces. Although LFDA, length-based techniques. This section outlines dating from an earlier epoch of computers, only three of these methods: the first the estimation of runs a maximum of 10 by 10 values of L• and k, yield-per-recruit in multispecies fishery using a analyses can be repeated over a series of values to single gear such as trawl (see also Shepherd and create very detailed GOF surfaces, and was used to Pope, Chapter 8, this volume); the second, a way of generate Plate 1. As far as I am aware, LFDA has not estimating the offshore migration rate, a common been recoded for Windows, but it runs effectively problem in many inshore tropical fisheries; and in a DOS window in Windows 98. the third, using a length-based analysis to estimate Using maximum-likelihood and automatic trophic level. direct computer search, the MIX pro- gramme (Ichthus Software, Hamilton, Ontario, 9.3.1 Ringing the changes on Canada; Macdonald and Green 1988; http:// Thomson and Bell: multispecies icarus.math.mcmaster.ca/peter/mix/mix31.html) yield-per-recruit finds a mixture of overlapping components which maximizes goodness-of-fit with the length– Fishing gear such as the mesh in a trawl selects fish frequency data. The distributions can be normal, by size, not age. Converting size to age for one log normal, exponential or gamma. A Windows species in the standard way means that yield- version is available. per-recruit (YPR) calculations comparing long- FAO distributes DOS software called FiSAT term average yield under a range of mesh sizes and (FAO-ICLARM Stock Assessment Tools; fishing rates can be made for this fishing gear. But if http://www.fao.org/fi/statist/fisoft/iclarm.asp) more than one species is caught in the trawl, the which implement an updated and improved same size of fish can represent different ages in dif- ELEFAN method, including the seasonal ferent species. This means that the yield for each growth model option and a plot of the GOF species must be calculated for each length class. surface. FiSAT also includes SLCA and the Once this is done, the yields can be added to pro- Bhattacharaya method. A Windows version is in vide multispecies yield-per-recruit estimate for development. the trawl gear. The most sophisticated computer package The basic length-based yield equation was available is MULTIFAN (Otter Software, Victoria, devised by Thomson and Bell, and is thoroughly BC, Canada. http://otter-rsch.com/index.html), described in Sparre and Venema (1992). A version initially published with an impressive example of this multispecies yield-per-recruit analysis is Size-based Methods in Fisheries Assessment 205

obtained from time series, or, as in this case, by Present running a Beverton and Holt biomass-per-recruit (BPR) analysis for each species at the current esti- mated F and age of entry zero. Relative recruit- ment indices were then obtained by using the estimated biomass of the species, since B/R = BPR and so R = B/BPR. Uncertainty (not shown here) was addressed using Monte Carlo simulations Multi-VPR 3.75 3 based on random sampling of 74 parameter distri- 2.25 butions. In this analysis yield-per-recruit has been 1.5 F 1 2 converted to value-per-recruit by multiplying each 3 4 0.75 5 6 7 species and size class by its market price. 8 0 9 Mesh size (cm) 10 The location of the present Hong Kong trawl fishery is indicated on the value-per-recruit sur- face of Fig. 9.8 by an arrow. The analysis suggests Present that, provided relative recruitment is not altered, the value of the existing fishery could be almost doubled by doubling the mesh size in the trawl (Pitcher et al. 1998).

9.3.2 Estimating the offshore Multi-VPR F migration rate using length and age 0 3 0.75 4 Often, fish migrate offshore into deeper water as 1.5 6 they get older and larger, partly as the appropriate 2.25 7 3 F 9 Mesh size (cm) refuge from predators changes and partly as a 3.75 10 search for larger food. By moving offshore, these may remove themselves from the sampling area, Fig. 9.8 Multi-species value-per-recruit (VPR) surface and this offshore migration rate may mask the true (front and back views) for the 17 Hong Kong species. VPR total mortality rate and decrease the apparent is plotted against mesh size in cm, and fishing mortality rate, F (vertical axis is value). Approximate present growth rate. If a sample of fish has been aged using position of fishery indicated by arrow. (Source: from otoliths or other hard parts, the differences be- Pitcher et al. 1998.) tween cohorts constructed from the biased growth parameters obtained from this method and those from the samples can be used to estimate the off- available in the FiSAT package, but the equa- shore migration rate with age (Pitcher et al. 1998). tions can fairly easily be programmed into a The method assumes that the probability of fish spreadsheet. migrating offshore out of the sampling zone An example of the results from a spreadsheet increases linearly with size from a start length length-based multispecies analysis applied to (P = 0.0) to an end length (P = 1.0) at which all fish Hong Kong trawl fisheries is shown in Fig. 9.8. The have left. work covered 17 species (14 fish and 3 inverte- First, the von Bertalanffy growth parameter, k, brates) caught in the Hong Kong trawl sector. The and L• is estimated from length–frequency analy- overall multispecies YPR was obtained by sum- sis and an auximetric plot that compares with pub- ming the contributions of each of the species, lished values. (For the lizardfish (Saurida tumbil) weighted by their relative recruitment, as advised in Fig. 9.9, L• = 69cm and k = 0.228.) Keeping the by Murawski (1984). Relative recruitment can be same L•, a curve fitted to otolith-derived ages gave 206 Chapter 9

1.0 Table 9.2 Estimated instantaneous rates of offshore migration for three species of Hong Kong fishes.

0.8 Age (year) L infinity 1.5 2.5 3.5

Lizardfish 69 0.90 3.70 – 0.6 Croaker 33 4.86 – 6.94 Ponyfish 14 1.85 1.88 –

0.4 Relative frequency Probability of migrating 0.2 start and end length, or an automatic search for a minimum performed. Examples of the goodness-of-fit surface for 0 6 1014182226303438 three Hong Kong fishes are shown in Plate 2 and the calculated instantaneous migration rates with Length (cm) age are shown in Table 9.2. The results show Fig. 9.9 Diagram showing the estimation of offshore that the small leiognathid ponyfish, which only migration rate with age for Saurida tumbil (lizardfish) reaches about 14cm maximum length, moves in Hong Kong. The analysis is performed on the first two offshore at about the same rate once adult at 1+. age groups. Start and end lengths for migration proba- The croaker moves rapidly out of the inshore area bility (thick inclined line) are adjusted until means and soon after it is a year old, while most lizard fish re- standard deviations of two cohorts (solid bell-shaped main inshore until their second year. curves) fit observed truncated distributions (broken curves). True mean lengths are 17.5 and 36cm, while the mean lengths of fish remaining are 14.1 and 19.1cm. 9.3.3 Estimating trophic levels using length and Ecopath Pauly et al. (2000) have developed a simple an apparent k of 0.96 year-1. Mean lengths at age method of calculating change in trophic level calculated from this apparent growth curve will be (DTL) where mean body sizes have been reduced by biased if fish migrate out of the sampling area. The fishing. The estimation depends on an empirical actual mean lengths at these ages is given by the relationship between body size and TL in 180 true growth curve. The next assumption is that species of fishes: the true cohorts have the same COV as the youngest cohort in the otolith samples (in this ex- DTL=+ blog10{[] L• {[ Lc() M k()1- E ample, 0.23). If this is true, then ‘ideal’ cohorts []()Mk()11- E+ }] with normally distributed lengths, including both []L• + {}() LcM k() M k + 1 , (9.15) unmigrated and migrated fish, can be set up using the mean lengths from the true growth curve. A where Lc = length at first capture; E = exploitation distribution of fish remaining is then set up by rate, F/Z; M = natural mortality rate; b = slope pf multiplying the ‘ideal’ cohort vector by the proba- relationship between TL and body length in cm bility of migrating for each length class. If the cor- which is 0.63 for large carnivores like cod, and 0.24 rect migration line is chosen, the mean length of for small pelagic and demersal fishes; and other pa- this distribution should be equal to the observed rameters are as above. mean length for the apparent growth curve. This method has been used in analysis of de- Goodness-of-fit is judged by least squares of the cline of TL in Canadian fisheries; the length-based two pairs of mean lengths and standard deviations. method provided very similar answers to an age- The analysis is repeated for all combinations of based method. Size-based Methods in Fisheries Assessment 207

Growth and age structure of the catch are es- 9.4 CONCLUSIONS timated simultaneously with population pa- rameters such as recruitment, selectivity, After over 120 years of length–frequency analysis catchability and natural mortality . . . we are justified in asking ‘Are we there yet?’, and Spatial structure can be included in the ‘Where do we go from here?’ model. . . . Missing data and data of different At one time, length-based methods were regard- temporal resolutions are allowable . . . ed as being suitable only for the tropics. This has Auxiliary data (such as tagging data) can be been exposed for the myth that it is (e.g. Pauly incorporated into the model as appropriate. 1994). Since that time, the discovery of daily Various structural hypotheses, such as otolith rings has meant that, at least for the young, density-dependent growth, trends ...and tropical fish can be aged. And length–frequency seasonal catchability, can be incorporated analysis has become a cost-effective way of mini- into the model and tested. mizing expensive ageing for many temperate fishes. An unexpected practical use for length– An example of Monte Carlo simulations exploring frequency analysis has arisen from Froese and the sensitivities of such a complex model is given Binholan’s (2000) technique, which has led to the by Fu and Quinn (2000). printing of posters allowing buyers at markets to In a different direction, length-based short reject fishes that are too small. cuts that can be programmed on a spreadsheet can For conventional length–frequency analysis to tackle a range of interesting problems – doubtless determine growth and mortality rates, it seems at many of these still remain to be invented and used. first sight that there is nowhere much left to go. But there may be completely new techniques on Parametric methods are now fully matured, and the horizon. One recent development may point to classic analyses such as mixture analysis may an exciting way to deal with our opening example be carried out, complete with uncertainty, on a of estimating the growth and mortality of humans spreadsheet. Among the non-parametric methods, massed in St Peter’s Square. Smith and Botsford most of which may also be performed on spread- (1998), who bravely say that it is ‘analytically sheets, the projection matrix stands out as being more challenging to analyze size frequency distri- deserving of further development and rigour. For butions that lack multiple age pulses’, show how most normal fisheries work in both the tropics and von Bertalanffy parameters estimated from tagging higher latitudes, in the past decade these analyses of individuals can be informative about the shapes have become robust and reliable standard tools. A of length–frequency distributions. Through over- word of caution is always to map the goodness-of- layed Monte Carlo simulations, which can very fit surface rather than rely on optimization rou- quickly be graphed on modern computers; they tines built into spreadsheets. A further warning is demonstrate how characteristic length–frequency always to perform several different analyses and shapes arise even in continuously recruiting compare the results. animals. The technique has been applied to fish In one direction, the research frontier of con- such as lingcod (Ophiodon elongatus), and inver- ventional length–frequency analysis is now mov- tebrates such as whelks (Buccinium undatum), ing into highly complex integrated catch-at-length surf clams (Macromeris polynyma), and urchins models that are far moved from analyses that most (Strongylocentrus franciscanus). fisheries scientists can perform on a spreadsheet. Long-lived, slow-growing fish that are difficult For example, MULTIFAN-CL (Fournier et al. 1998) to age and assess using conventional means is described as a length-based, age-structured, like- currently remain intractable even with complex lihood model that circumvents many of the diffi- length-based methods. This problem has not culties associated with sequential analyses such as changed. But it is a gap in the fisheries assessment VPA (Shepherd and Pope, Chapter 7, this volume): toolbox now that serial depletion by area, depth, 208 Chapter 9 species and taxon proceeds apace in the world’s Froese, R. and Binholan, C. (2000) Empirical relation- overexploited oceans. It seems as though the criti- ships to estimate asymptotic length, length at first cal problems for the fisheries of the early 21st cen- maturity and length at maximum yield per recruit in fishes, with a simple method to evaluate length fre- tury have bypassed single-species analyses such as quency data. Journal of Fish Biology 56, 758–73. those presented here. Are length-based ecosystem Fu, C. and Quinn, T.J. (2000) Estimability of natural mor- analyses the hope of the future? tality and other population parameters in a length- based model: Pandalus borealis in Kachemak Bay, Alaska. Canadian Journal of Fisheries and Aquatic ACKNOWLEDGEMENTS Sciences 57, 2420–32. Gayanilo, F.C., Sparre, P. and Pauly, D. (1996) FAO- ICLARM Stock Assessment Tools (FiSAT): User Man- I am grateful to Peter Macdonald who, a long time ual. Rome: FAO. ago, initiated me into formal length–frequency Gulland, J.A. (1983) Fish Stock Assessment: A Manual of analysis using what was then a revolutionary new Methods. London: Wiley. 4K computer, the Elliot 803. I also thank Daniel Gulland, J.A. and Rosenberg, A.A. (1990) A Review of Pauly, who, not so long ago, provided critical Length-based Approaches to Assessing Fish Stocks. comments on a draft of this manuscript. FAO Fisheries Technical Paper, No. 323. Rome: FAO. Hald, A. (1952) Statistical Theory with Engineering Applications. New York: Wiley. Harding, J.P. (1949) The use of probability paper for graphi- REFERENCES cal analysis of polymodal frequency distributions. Journal of the Marine Biological Association UK 28, Basson, M., Rosenberg, A.A. and Beddington, J.R. 141–53. (1988) The accuracy and reliability of two new meth- Harris, D. (1968) A method of separating two superim- ods for estimating growth parameters from length- posed normal distributions using arithmetic proba- frequency data. Journal du Conseil International pour bility paper. Journal of Animal Ecology 37, 315–19. L’Exploration de la Mer 44, 277–85. Hilborn, R. and Mangel, M. (1997) The Ecological Detec- Behboodian, J. (1970) On the modes of a mixture of two tive: Confronting Models with Data. Princeton, NJ: normal distributions. Technometrics 12, 131–9. Princeton University Press. Bhattacharya, C.G. (1967) A simple method of resolution Isaac, V.J. (1990) The accuracy of some length-based of a distribution into Gaussian components. Biomet- methods for population studies. ICLARM Technical rics 23, 115–35. Report, No. 27. Manila, Philippines: ICLARM. Buchanan-Wollaston, H.G. and Hodgeson, W.C. (1929) A Kerstan, M. (1995) Ages and growth rates of Agulhas new method of treating frequency curves in fishery sta- Bank horse mackerel Trachurus trachurus capensis: tistics with some results. Journal du Conseil Interna- comparison of otolith ageing and length frequency tional pour L’Exploration de la Mer 4, 207–25. analyses. South African Journal of Marine Science 15, Cassie, R.M. (1954) Some uses of probability paper in the 137–56. analysis of size frequency distributions. Australian Kleiber, P. and Pauly, D. (1991) Graphical representations Journal of Marine Freshwater Research 5, 513–22. of ELEFAN 1 response surfaces. Fishbyte 9 (2), 45–9. Everitt, B.S. and Hand, D.J. (1981) Finite Mixture Distri- Koranteng, K.A. and Pitcher, T.J. (1987) Population para- butions. London: Chapman & Hall. meters, biannual cohorts and assessment in the Pagel- Fournier, D.A., Sibert, J.R., Majkowski, J. and Hampton, lus bellottii (Sparideae) fishery off Ghana. Journal du J. (1990) MULTIFAN: a likelihood-based method for es- Conseil International pour L’Exploration de la Mer 43, timating growth parameters and age composition from 129–38. multiple length frequency data sets illustrated using LeCren, E.D. (1992) Exceptionally big individual perch data for southern bluefin tuna (Thunnus maccoyii). (Perca fluviatilis L.) and their growth. Journal of Fish Canadian Journal of Fisheries and Aquatic Sciences Biology 40, 599–625. 47, 301–17. Liu, Q., Pitcher, T.J. and Al-Hossaini, M. (1989) Ageing Fournier, D.A., Hampton, J. and Sibert, J.R. (1998) with fisheries length–frequency data using informa- MULTIFAN-CL: a length-based, age-structured model tion about growth. Journal of Fish Biology 35 (A), for fisheries stock assessment, with application to 167–77. South Pacific albacore, Thunnus alalunga. Canadian Macdonald, P.D.M. and Green, P.E.J. (1988) User’s Guide Journal of Fisheries and Aquatic Sciences 55, 2105–16. to Program MIX: An Interactive Program for Fitting Size-based Methods in Fisheries Assessment 209

Distribution Mixtures. Hamilton, Ontario: Ichthus Pitcher, T.J., Watson, R., Courtney, A.M. and Pauly, D. Data Systems. (1998) Assessment of Hong Kong’s inshore fishery Macdonald, P.D.M. and Pitcher, T.J. (1979) Age-groups resources. Fisheries Centre Research Reports 6 (1), from size-frequency data: a versatile and efficient 1–168. method of analyzing distribution mixtures. Journal of Rosenberg, A.A. and Beddington, J.R. (1987) Monte Carlo the Fisheries Research Board of Canada 36, 987–1001. testing of two methods for estimating growth from Murawski, S.A. (1984) Mixed species yield-per- length–frequency data, with general conditions for recruitment analyses accounting for technological their applicability. In: D. Pauly and G.P. Morgan interactions. Canadian Journal of Fisheries and (eds) Length-based Methods in Fisheries Research. Aquatic Sciences 41, 897–916. ICLARM Conference Proceedings, No. 13. Manila, Pauly, D. (1984) Fish population dynamics in tropical wa- Philippines: ICLARM, pp. 283–98. ters: a practical manual for programmable calculators. Rosenberg, A.A. and Beddington, J.R. (1988) Length- Manila, Philippines: ICLARM. based methods of stock assessment. In: J.A. Gulland Pauly, D. (1987) A review of the ELEFAN system for (ed.) Fish Population Dynamics, 2nd edn. New York: analysis of length–frequency data in fish and aquatic Wiley, pp. 83–103. invertebrates. In: D. Pauly and G.P. Morgan (eds) Rosenberg, A.A., Beddington, J.R. and Basson, M. (1986) Length-based Methods in Fisheries Research. The growth and longevity of krill during the first ICLARM, Conference Proceedings, 13. Manila, Philip- decade of pelagic whaling. Nature 324, 152–4. pines: ICLARM, pp. 7–34. Schnute, J. and Fournier, D. (1980) A new approach to Pauly, D. (1994) May apply in the tropics, but not here! length–frequency analysis: growth structure. Journal In: On the sex of Fish and the Gender of Scientists. of the Fisheries Research Board of Canada 37, London: Chapman & Hall, pp. 3–7. 1337–51. Pauly, D. and David, N. (1980) An objective method Shepherd, J.G. (1987a) A weakly parametric method for for determining growth from length–frequency data. the analysis of length composition data. In: D. Pauly ICLARM Newsletter 3 (3), 13–15. and G.P. Morgan (eds) Length-based Methods in Fish- Pauly, D. and Munro, J.L. (1984) A simple method for eries Research. ICLARM Conference Proceedings, No. comparing the growth of fish and invertebrates. Fish- 13. Manila, Philippines: ICLARM, pp. 113–20. byte 2 (1), 21. Shepherd, J.G. (1987b) Towards a method for short-term Pauly, D. and Caddy, J. (1985) A Modification of Bhat- forecasting of catch-rates based on length composi- tacharya’s Method for the Analysis of Mixtures of Nor- tion. In: D. Pauly and G.P. Morgan (eds) Length-based mal Distributions. FAO Fisheries Circular, No. 781. Methods in Fisheries Research. ICLARM Conference Rome: FAO. Proceedings, No. 13, Manila, Philippines: ICLARM, Pauly, D. and Arreguin-Sanchez, F. (1995) Improving pp. 167–76. Shepherd’s length composition analysis (SLCA) for Smith, B.D. and Botsford, L.W. (1998) Interpretation of growth estimations. Naga, the ICLARM Quarterly 18 growth, mortality and recruitment patterns in size- (4), 31–3. at-age, growth increment, and size frequency data. Pauly, D., Palomares, M.L., Froese, R., Pascualita, S., Canadian Special Publication in Fisheries and Aquat- Vakily, M., Preikshot, D. and Wallace, S. (2000) Fishing ic Sciences 125, 125–39. down Canadian aquatic food webs. Canadian Journal Smith, B.D., Botsford, L.W. and Wing, S.R. (1998) Estima- of Fisheries and Aquatic Sciences 58, 51–62. tion of growth and mortality parameters from size fre- Petersen, C.G.J. (1891) Eine Methode zur Bestimmung quency distributions lacking age patterns: the red sea des Alters und Wuchses der Fische. Mitteilungen des urchin as an example. Canadian Journal of Fisheries Deutschen S’eefischerei-Verbandes 11, 226–35. and Aquatic Sciences 55, 1236–47. Pitcher, T.J. (1971) Population Dynamics and Schooling Sparre, P. and Venema, S.C. (1992) Introduction to Behaviour in the Minnow, Phoxinus phoxinus (L.) D. Tropical Fish Stock Assessment, vol. I: Manual. Phil thesis, Oxford University, 377pp. FAO Fisheries Technical Paper, No. 306 (1). Rome: Pitcher, T.J. and Macdonald, P.D.M. (1973) An integra- FAO. tion method for fish population fecundity. Journal of Tanaka, S. (1962) A method of analysis of a polymodal fre- Fish Biology 4, 549–53. quency distribution and its application to the length Pitcher, T.J., Bundy, A. and Neill, W.S. (1996) The fishery distribution of porgy, Taius tumifrons. Journal of the for Rastrineobola argentea in Lake Victoria: estima- Fisheries Research Board of Canada 19, 1143–59. tion of potential yields using a new approximate Taylor, B.J.R. (1965) The analysis of polymodal frequency method based on primary production. Fisheries Re- distributions. Journal of Animal Ecology 34, 445– search 28, 133–49. 52. 210 Chapter 9

Terceiro, M. and Idoine, J.S. (1990) A practical assess- movements and diet of the terapontid Amniataba ment of the performance of Shepherd’s length compo- caudavittata in an Australian estuary. Journal of Fish sition analysis (SRLCA): application to Gulf of Maine Biology 45 (6), 917–31. Northern Shrimp Pandalus borealis survey data. Fish- Yakowitz, S.J. (1969) A consistent estimator for the iden- ery Bulletin 88, 761–73. tification of finite mixtures. Ann. Math. Stat. 40, Wise, B.S., Potter, I.C. and Wallace, J.H. (1994) Growth, 1728–35. 10 Ecosystem Models

DANIEL PAULY AND VILLY CHRISTENSEN

10.1 INTRODUCTION: when fishing that it destroys the flowers of ECOSYSTEM MODELS the land below water there, and also spat of oysters, mussels and other fish upon which Historical sources indicate that the fishers and the great fish are accustomed to be fed and scholars of centuries past were well aware that the nourished. (Alward 1932) fish and invertebrates then exploited by fisheries were embedded within what we now call ‘ecosys- Incorporating the commonsense knowledge tems’ (see e.g. Thompson 1910; Cotte 1944). As expressed here into a rigorous ecosystem-based well, this knowledge was used to draw inferences understanding of fisheries took, however, an ex- structurally similar to contemporary hypotheses. tremely long time, and the path taken was particu- Thus, for example, Bernard de Palissy (1510–90) in- larly tortuous, with various blind alleys and terpreted fossil fish as the remnants of populations enticing traps (see also Smith, Chapter 4, this that had been ‘trop pêchées’ (i.e. overfished to volume). Indeed, many fisheries scientists, accus- extinction; Tort 1996, p. 3345), thus admitting to tomed to ‘single-species stock assessments’, are the possibility of humans having such impacts – still uncomfortable with the broad ecological prin- possibly based on observations at the time. This ciples required for such understanding, not to inference that fishing had ecosystem effects is speak of regulatory agencies, whose mandate often illustrated by a petition from the year 1376 to precludes them from even beginning to address Edward III, King of England: ecosystem-related issues (NRC 1999). This chapter therefore introduces a number of The commons petition the King, complain- ecological principles which formalize and render ing that where in creeks and havens of the mutually compatible earlier ideas on how the dif- sea there used to be plenteous fishing, to the ferent elements of aquatic ecosystems interact. profit of the Kingdom, certain fishermen for Moreover, it illustrates how applying these princi- several years past have subtly contrived an ples can improve one’s understanding of a given instrument called ‘wondyrechaun’ made in aquatic ecosystem, and of such systems in general. the manner of an oyster dredge, but which is The link with fisheries is then established by considerably longer, upon which instrument documenting how fish and invertebrate resource is attached a net so close meshed that no fish, species are embedded in ecosystems and how they be it ever so small, that enters therein can es- must (and do) follow the same principles as non- cape, but must stay and be taken. And that exploited species. Our goal, thus, is to marry key the great and long iron of the wondyrechaun parts of Ecology with the best parts of Fisheries runs so heavily and hardly over the ground Science. In doing so, we build on the foundations of 212 Chapter 10 the two chapters in Volume 1 that examine trophic tion taken up by the few aquatic insects that make interactions in freshwater (Persson, Chapter 15) it to the adult stage, and by birds, which is available and marine communities (Polunin and Pinnegar, for transfer to adjacent, terrestrial ecosystems. Chapter 14). Our chapter, in turn, provides a theo- This definition, however, also fits more open retical and methodological underpinning for the systems, such as coral reefs, seamounts or estu- ecosystem impacts of fishing documented by aries, whose different components have co-evolved Kaiser and Jennings (Chapter 16, this volume). such that most of their trophic fluxes are ‘inter- Both ecology and fisheries science share the nalized’, if at the cost of large supporting energy same ancestry, having been spun off Natural fluxes, in the form of light for coral reefs (Grigg History in the late 1800s. However, their subse- 1982), zooplankton for seamounts (Koslow 1997), quent fates differed. Ecology went to University, or detritus for estuaries (Odum and Heald 1975). On theorized, and usually failed to get its hands dirty. the other hand, this definition of ecosystems may Fisheries science, on the other hand, decided to exclude beaches and other shallow-water areas work for the Government, and became very ap- whose inhabitants often consist of the juvenile plied, hence a prime suspect every time some fish form of adults living in adjacent, deeper waters. stock collapses. The union we advocate may be The concept of ‘subsystems’ is appropriate for such thus likened to a shotgun marriage: the partners do highly connected parts. As will be shown below, not necessarily match, but strong external con- stratification by subsystems is crucial when con- straints force them together. The shotgun itself is structing models of ecosystems, our next topic. the crisis of fisheries, and the ensuing threats to Models are coherent representations of systems the systems studied by both (aquatic) ecologists and/or of the processes therein, and may consist of and fisheries scientists, and as well, to the funding words (‘word models’), graphs or equations. Words of these often feuding disciplines. A similar con- alone can often describe complex systems or vergence of protected area management and fish- processes adequately, as in the case of natural eries has been identified in the context of marine selection (Darwin 1859). Graphs can also make protected areas by Polunin (Chapter 14, Volume 1). compelling cases, as did, for example, the trophic Before all, however, we must define the con- pyramid of Lindeman (1942). However, equations cepts in our title, ‘ecosystems’ and ‘models.’ that capture essential aspects of systems or pro- Ecosystems, in this chapter are sites occupied by cesses always outperform word or graph models, elements such as individual members of different if only because the application of standard plant and/or animal species, which interact such algebraic or other mathematical rules to these that the sum of their mutual interactions is much equations often leads to the discovery of unknown greater than the sum of their interactions with the properties of the systems or processes in question. elements of adjacent sites. The interactions within This non-intuitive, and in fact wondrous property the ecosystems thus defined may lead to distinct of mathematical descriptions (Wigner 1960) has, structures for these ecosystems. However, our moreover, the distinct advantage of allowing the definition does not require adjacent ecosystems to testing of hypotheses about future states, or previ- be structurally different from each other. ously unobserved features of ecosystems, besides Lakes are excellent examples of ecosystems de- allowing for testing the adequacy of the initial fined here (Thienemann 1925; Golley 1993). Thus, description. if consumed at all, the primary production of a lake Our models will thus be mainly equations, from phytoplankton and macrophytes is grazed though not necessarily complex ones. The criteri- within that lake, as is its secondary production on for assessing the value of a model is not the ex- which is composed of zooplankton, and the pro- tent to which it reproduces the complexity of the duction by fish and other organisms is supported real world. Rather, it is the ratio of insight gained by feeding on the secondary producers. Indeed, in versus effort extended. As we shall see, ‘simple’ lakes, it is usually only that part of primary produc- models that capture key features of an ecosystem Ecosystem Models 213 in an obvious fashion are far more useful than com- yield to a fishery depends on the character and plex, but opaque models. Sparre and Hart (Chapter strength of these interactions (Kirkwood 1982). 13, this volume) review both philosophical and Thus, the combined yield of a prey and predator is practical aspects of using models. always less than these organisms considered sepa- rately. On the other hand, species with mutually beneficial interactions increase the combined 10.2 MULTISPECIES yield. These results seem trivial; however, we MODELS must realize that single-species fisheries models still largely assume that predation effects can be The first formal representation of interactions be- neglected, with optima being proposed separately tween predator and prey were published by Lotka for each species, as if they did not interact. On the (1925) and Volterra (1926) and the system of equa- other hand, the familiarity that marine scientists tions they proposed to describe such interactions, and ecosystem modellers now have with the now known as Lotka–Volterra equations, are still Lotka–Volterra approach has had numerous nega- used – often in strongly modified forms – for ex- tive impacts. Notably, modellers have uncritically ploring mathematically the consequences of cer- taken over the mass-action assumption implied in tain features of systems of interacting organisms the Lotka–Volterra model, i.e. that the consump- (see Smith, Chapter 4, this volume). The basic tion of prey by predators is a function of, and only form of this system is: of, their respective numbers or biomass in the ecosystem, just as is the case with reactants in a ddNt1121=-() rcNN (10.1a) chemical vat. As we shall see, this assumption ignores the use of spatial structure by potential ddNt2212=-() gcNN + , (10.1b) prey to enable part of their population to hide from, or otherwise render itself invulnerable to, where r is the intrinsic rate of population growth of their predator(s) (Krause et al., Chapter 13, Volume the prey, g is the coefficient of negative growth (de- 1). This dampens the oscillation otherwise im- cline) of the predators (N2) in the absence of prey plied by the system. Spatial structure also enables (N1) and c1, c2 are interaction coefficients. Given the coexistence of two very similar competing certain sets of values for these parameters, the species, whereas the equation systems such as predator and the prey will oscillate, sometimes equation (10.1) would predict that one would have violently: the prey grows, the predator follows to go extinct (Gause 1934). and over-consumes the prey which declines, fol- Another more practical approach to analysis of lowed by the predator with the prey then picking multispecies situations is to add up yield-per- up again. Volterra, after modifying these equations recruit (i.e. the results of yield-per-recruit analy- such as to cover any number of species, teamed up ses, see Shepherd and Pope, Chapter 8, this with his son-in-law, the fisheries scientist Umber- volume) for a number of species into an overall to D’Ancona, to apply the new formalism to the in- yield-per-recruit for a given area. Multispecies terpretation of observed cycles in the landings of yield-per-recruit analysis, while tedious to imple- small prey and large predatory fishes from the Adri- ment, is conceptually straightforward (Beverton atic. The details of this story are in Tort 1996 and and Holt 1957; Munro 1983; Murawski et al. 1991). the historical setting is given by Smith (Chapter 4, However, the results suffer from the same defect this volume). We note in passing that such cycles, that plagues single-species yield-per-recruit analy- nowadays, would invariably be (if often unjustifi- ses, that is, strong assumptions must be made as to ably) attributed to environmental fluctuations. the level of recruitment for each of the species con- The exploration of systems of equations such as sidered. Usually, this level of recruitment is as- (10.1a and b) has led to a number of insights about sumed constant, and the solutions, which then interacting species, notably, that their combined rely on equilibrium assumptions, ignore the vio- 214 Chapter 10 lent transient effects often resulting from fishing other, each subjected to a different stage-specific mortality changes (Shepherd and Pope, Chapter 8, mortality, and the vagaries of a varying physi- this volume). However, when the signal from the cal environment. Still, the aggregate behaviour fishery being investigated is very strong, such as in of the model was credible; notably, it generated fisheries in which a wide range of species is being a particle-size distribution such as observed in- growth overfished, this approach can provide esti- dependently (see Section 10.8 below), and natural mates of the gain that can be obtained, by, say, re- mortalities for juvenile fishes well in excess of the ducing fishing mortality on a wide range of species. values then assumed, for example about M = 2 In such cases, indeed, the method is particularly year-1 instead of about 0.1 year-1 for juvenile her- powerful in that it allows one to distinguish be- rings. This result was later confirmed by multi- tween fisheries with different selection patterns. species virtual population analysis, MSVPA (see It also allows one to identify, again under the Pope and Shepherd, Chapter 7, this volume). assumptions of constant recruitment and of no However, it is clear that few of Andersen and biological interactions, the mix of fishing effort Ursin’s colleagues were convinced of the practical- by size or age which optimizes overall yield-per- ity of this tool for fisheries management. Rather, recruit. Thus, in spite of its clear drawbacks, these what it established was that simulating a large techniques are likely to continue being used, spe- marine ecosystem was not an intractable task. cially since they are not very demanding in terms Moreover, several of the elements of the Andersen of input parameters. These are growth parameters, and Ursin model – notably its size-selection sub- estimates of natural mortality, and others, most of model (Ursin 1973) – were found to be extremely which, at least for commercial species, can be ob- useful on their own, and have seen use in tained online; see www.fishbase.org. One software later modelling approaches, e.g. the length-based system that can be used to implement this MSVPA described by Christensen (1995a) and techniqueisFiSAT(Gayaniloetal.1996),also avail- Shepherd and Pope, Chapter 7, this volume. able online (www.fao.org/fi/statist/fisoft/fisat/ The simulation of the Northeastern Pacific project.htm). ecosystem developed by Laevastu and collabora- tors, though similarly ambitious, differed from 10.3 MODELLING THE Andersen and Ursin’s approach in that no attempts were made to track numbers of individuals. NORTH SEA, THE NORTH Rather, the model strived to maintain mass bal- PACIFIC AND THE GULF ance between the cells of a spatial grid, upon which OF THAILAND were overlaid the biomass of commercial fishes, and of their prey called ‘forage fishes’, and preda- Simulation models, incorporating basic elements tors which were mostly marine mammals. Here of the Lotka–Volterra approach, but also incorpo- again, the model and its predictions were not rating components such as those used in yield-per- adopted as such. Rather, the PROBUB model, as it recruit analyses, may be seen as the next logical was later called (Laevastu and Larkins 1981), at- step towards dealing with multispecies interac- tracted the attention of researchers in need of a tool tions. Indeed, several attempts have been made to for modelling biomass flows in a data-sparse sys- conceptualize and parameterize simulation mod- tem, the French Frigate Shoals, Northern Hawai’i els of large marine ecosystems, and three of these (Polovina and Tagami 1980). Stripped of its spatial had a great indirect and direct impact on fisheries and temporal dynamics, PROBUB led to the first research. The North Sea model of Andersen and version of Ecopath, a simple mass-balance ap- Ursin (1977) is unique in terms of its extreme com- proach for estimating biomasses in aquatic sys- plexity and level of details, the consequence of an tems (Polovina 1984). Another indirect effect of attempt to track the numbers of fish from the eggs work on PROBUB is that it motivated a thorough to the adults of dozens of species feeding on each study of the diet of Bering Sea fishes, and thus en- Ecosystem Models 215 abled the construction of successor models (Trites tion, should be that used for determining one’s use et al. 1999; also see below). of one or the other formulation. Polovina (1984), Another early simulation model relevant to who developed the original Ecopath formulation, fisheries was that of Larkin and Gazey (1982), de- emphasized the biomasses as the unknowns in his scribing the Gulf of Thailand. Like Andersen and system, thereby making strong assumptions about Ursin for the North Sea and Laevastu and collabo- the fraction of the fluxes retained within the sys- rators for the North Pacific, Larkin and Gazey were tem. In the reformulation of the Ecopath approach faced with the need to parameterize a large number (Christensen and Pauly 1992; Pauly et al. 1993), of coupled differential equations. However, con- this emphasis on biomass was replaced by an op- trary to the situation for the North Sea and the tion to leave other elements of a system of fluxes Pacific, there was, at the time, no hope of obtain- unknown, notably the ecotrophic efficiency (EE), ing the parameters required for a model of the Gulf expressing the fraction of mortality not due to pre- of Thailand, then much less studied than the tem- dation or fishing, a parameter far more difficult perate North Sea, or the North Pacific. Thus they than biomass to estimate in the field. pushed as far as they could the process wherein evo- The well-known master equation for Ecopath lutionary first principles are used to derive assump- is, for each functional group i: tions about the likely parameter that such values might take. The result, while establishing their in- BPBEEY BQBDC, (10.2) i ◊ ()i ◊=+◊ii j()j ◊ ij genuity, confirmed that simulation models struc- tured around coupled differential equations could where Bi and Bj are biomasses (the latter pertaining not be parameterized for most situations such that to j, the consumers of i); P/Bi their production/bio- fisheries scientists and managers would feel com- mass ratio, equivalent to total mortality under fortable with the results. Clearly, a route had to be most circumstances (Allen 1971); EEi the fraction found to harness the scattered data on catches, of production (P = B◊(P/B)) that is consumed with- standing stock estimates, and food consumption in, or caught from the system (usually left as the rates, which marine and fishery biologists have unknown to be estimated when solving the equa- published over the years but which could not be tion above); Yi is the fisheries catch (i.e. Y = F◊B); straightforwardly used to parameterize models Q/Bj the food consumption per unit biomass of j; such as the one of Larkin and Gazey. This, again, and DCij the contribution of i to the diet of j. leads to the Ecopath model, our next topic. Herein, solutions for the unknowns, e.g. Bi, are obtained by solving the matrix system in equation (10.2) through a robust inversion routine (MacKay 10.4 ECOPATH AND THE 1981). (The right-hand side of equation (10.2) can

MASS-BALANCE APPROACH also include a biomass accumulation term (Bacc) in cases where the biomass is known to have changed The basic idea behind the mass-balance approach during the period under consideration, thus allow- incorporated in Ecopath is both trivial and pro- ing for non-equilibrium situations (see below), as found. At any time within the system, and within well as a term for net migration.) the elements of that system, the amounts of matter Solving the system of equations (10.2) does not that flow in must balance the amount that goes out require the Ecopath software per se. Indeed, they plus the change in biomass. This means that if can be implemented on spreadsheets (see e.g. parts of the fluxes in a system and/or biomass in a Mathisen and Sands 1999). Rather, their key at- system are known, the values for the other parts are tribute is that it is easy to find estimates of various constrained and therefore can be estimated by processes and states in ecosystems which corre- subtraction. This principle can be implemented spond to the parameters of these equations, thus through different sets of equations and we think largely solving the parameterization problem that that the utility criterion, presented in the Introduc- has been besetting earlier modelling approaches. 216 Chapter 10

Moreover, to correct a false impression created (see models in Christensen and Pauly 1993). Pre- by the title of Christensen and Pauly (1992), the sent applications typically include more groups, Ecopath master equations do not require equilibri- from 20 to 50, as required, to represent a wide range um or steady-state. Rather, they require that mass of ecological and spatial variability and ontogenic balance occurs during the period under considera- diet changes. Whether one uses few or many state tion, e.g. that any excess consumption during part variables, it is important that all taxonomic groups of that period is compensated for by reduced con- occurring in a system be included, whether or not sumption during another part. This is particularly detailed data on them are available. This can be important for models of ecosystems that included done by including a group explicitly, with its own a strong seasonal component and which either can state variables. This would be for common, well- be represented by parameters averaged over a year, documented species. Alternatively it can be done which is the option chosen for many applications implicitly as part of broadly defined functional so far, or in which monthly change in biomass, diet groups. A detritus group must also be included, to composition, Q/B and P/B are modelled explicitly. which all fluxes of ungrazed plants and/or uncon- Constructing an Ecopath model is thus a matter sumed dead animals are directed. Bacteria may or of performing the following steps: may not be added as an explicit group. If they are 1 Defining the area in which the ecosystem oc- not an explicit group, it is assumed that they feed curs. This should preferably be one in which a dis- from the detritus. In this case, the EE of detritus tinct community of organisms occurs with limited must be <1; this implies that the detritus produced exchange to adjacent ecosystem. This system area in a system is not all consumed within that sys- may comprise several subsystems, e.g. a gulf may tem, i.e. this allows for consumption by bacteria consist of a shallow, mangrove-dominated area, a considered to reside ‘outside’ of that system. rocky or coralline intermediate area and deeper 4 Assembling available estimates of biomass, mud-covered grounds (Opitz 1993). This spatial ar- P/B, Q/B and diet composition. Contrary to the ticulation of a system in the form of subsystems is almost ritually invoked phrase that ‘nothing is important for the spatial considerations discussed known of x’, over a hundred years of quantitative further below. work by marine biologists, limnologists, fishery 2 Defining the period represented by an ecosys- scientists and others have generated a vast archive tem. Typically, this will be a period during which of valuable information on the aquatic ecosystems major field surveys have been conducted, provid- of the world. Much of it is, admittedly, scattered in ing estimates of biomass, diet composition, and a vast and still largely untapped literature, some of other important elements for many of the most it obscure. Constructing Ecopath models requires important groups in the system. This period is also access to that literature. This task is facilitated in one during which major changes in the biomass of part by FishBase, which can be made to output P/B, the major ecosystem elements can be assumed not Q/B and diet compositions for the thousands of to have occurred. Thus, wherever massive changes species for which such information has been en- are known to have occurred, it is better to repre- coded, and also provides a starting point for the lit- sent that system by models for each of the periods, erature search in question (Froese and Pauly 2000). before and after the change. Having at least two Only after such a literature search has been con- models also has the advantage that additional in- ducted should field programs be launched aiming ferences on the vulnerability of various species to at estimating missing parameters. Moreover, to their predators and to the fisheries can then be guide both the literature searches and the field made using Ecosim, presented below. work, routines are available in Ecopath which 3 Defining the state variables explicitly included quantify the uncertainty associated with the avail- in the system. Earlier applications of Ecopath able input values and thus help identify those tended to be composed of 10 to 20 functional components of a model requiring additional field groups representing all species in the ecosystem inputs. Ecosystem Models 217

Balancing a model then consists of solving the ogists worked ‘for’ rather than ‘with’ the few mod- system of linear equations (10.2), under the con- ellers then working on ecosystems (Golley 1993). straint that the EE values remain equal or less than This transition has markedly changed the field and 1, and that the respiration terms are all positive. made ecosystem-based management a realistic op- This balancing often involves revising initial esti- tion, as we shall show below, at least as a research- mates and is guided by the quantification of the un- able issue, if not in actual implementation. certainties as outlined in (4). Formal approaches for explicitly considering these uncertainties in- volve: (a) the Monte Carlo routine (‘EcoRanger’), 10.5 THE TROPHIC which uses the uncertainties in (4) in Bayesian LEVEL CONCEPT mode, as prior distributions, and outputs both dis- tributions of estimates, and posterior distributions Given a Darwinian world, the main limiting factor for the input values (Pauly et al. 2000, Fig. 3); and for most animals is ‘safe food’. This is, food that (b) a recently developed approach based on simu- can be consumed without exposing oneself unduly lated annealing, in which fuzzy logic is used to to predation (Jones 1982; Walters and Juanes 1993). modify qualitative diet compositions within con- Thus, modelling ecosystems is mainly a matter straints, until the system of equations is balanced. of describing food consumption by prey animal Once they are balanced and their key features (Mittelbach, Chapter 11, Volume 1; Juanes et al., checked for internal consistency and compati- Chapter 12, Volume 1), and the predation they are bility with similar ecosystems, models can be exposed to (Krause et al., Chapter 13, Volume 1). represented in the form of flow charts, which This is done either by tracking individuals and summarize these key features. Such flow charts their fate, or by tracking the fate of the biomass (Fig. 10.1) can be extremely information-rich and, incorporated in these individuals. The former ap- in the following, we discuss some aspect of this in- proach leads to single-species population dynam- formation in terms of trophic levels that are calcu- ics, well covered in this volume (see Sparre and lated and not previously assumed. This was also Hart, Chapter 13, this volume; Shepherd and Pope, the case with the Lindeman pyramids earlier used Chapters 7 and 8, this volume); while it is the latter to represent ecosystems. We then discuss other as- approach, involving biomass, or energy, a close pects of the information contained in ecosystem correlate (MacDonald and Green 1983) that is flow charts. emphasized in this chapter. We stress that the four steps outlined above Probably the simplest way to describe an ecosys- have now been implemented in hundreds of loca- tem is to re-express it as a ‘food chain’, separating tions, resulting in over 150 published models. its overall biomass into primary producers, herbi- These describe a vast range of ecosystem types vores, or first-order consumers, first-order carni- from data-rich areas, such as the North Sea and the vores or second-order consumers, etc., each of North Pacific, partly using, for both areas, the these links representing a trophic level (Lindeman same database used for the simulation models dis- 1942, Polunin and Pinnegar, Chapter 14, Volume cussed above, to the more data-sparse Gulf of Thai- 1). The International Biological Programme land (Christensen 1998) where the straightforward (Golley 1993) was structured around such integer data requirements of the Ecopath model enabled trophic levels, with field workers collecting data the parameterization based on empirical data that on biomass and production at each trophic level, had eluded Larkin and Gazey (1982). This vast and sending them to ‘modellers’ tasked with incor- range of applications demonstrates the applicabili- porating these data into simulation models. We ty of Ecopath as perceived by users who feel have mentioned above some of the disadvantages thereby empowered, contrary to the situation pre- of this separation of data collection and modelling vailing in programmes such as the International that this arrangement implied. It is, thus, sufficient Biological Programme (IBP) where many field biol- to add here the radical critique of the trophic 218 Chapter 10

5

Sharks Costa tuna Lg. piscivores

Scomberomorus Dem. piscivores Lutianidae Saurida 4 False trevally Mammals Jacks Pomfret Scianidae Priacanthus Dem. Sillago Cephalopods benthivores Plectorhynchus Catfishes Sm. Juv. sm. Rays demersals Juv. Nemipterus Juv. jacks pelagics Sm. Rastrelliger nemipterus 3 pelagics Ponyfishes Juv. Jellyfish Trophic level Crabs, lobsters 'Trash fish' Saurida

Penaeids Sergestids Benthos Misc. shellfishes Sea cucumbers Zooplankton 2

Detritus Seaweeds 1 Phytoplankton

Fig. 10.1 Flow diagram for a trophic model of the Gulf of Thailand. All flows between ecosystem groups are quantified, but only flows exceeding 5t·km2·year-1 are indicated. The sizes of the boxes are a function of the group biomasses. level concept which Rigler (1975) presented at a Heald 1975), computed, for the animals of a given key meeting of the IBP, based on the data collected population (i), from the trophic levels of all their by the IBP, showing that most aquatic animals, prey (j). Thus, we have: which include all those that are not strict her- bivores or detritivores, feed, simultaneously, at n TLijij=+1 Â TL ◊ DC . (10.3) different trophic levels. Rigler’s critique was ex- j=1 tremely influential, and its echoes are still de- tectable (see Cousins 1985). However, the solution where DCij is the fraction of j in the diet of i, and n to issues raised by this critique had a simple is the number of prey types. Estimates of trophic answer: fractional trophic levels (TLi; Odum and levels derived from equation (10.3) and earlier esti- Ecosystem Models 219

mates of TLj exist for marine mammals (Pauly et similar to the size or metabolic rate of organisms, al. 1998a), and for sharks (Cortés 1999) and other which can be measured by different, indepen- fishes (see Froese and Pauly 2000, and updates in dent methods, and whose various features, www.fishbase.org). therefore, can be elements of testable, quantitative Another widely used method to estimate frac- hypotheses. tional trophic levels is through the analysis of sta- ble isotopes of nitrogen (reviewed by Polunin and Pinnegar, Chapter 14, Volume 1). This relies on the observation that the ratio of 15N to 14N increases 10.6 PRACTICAL USES OF by about 3.4% every time proteins are ingested by a TROPHIC LEVELS: TRACKING consumer, broken down and resynthesized into its FOOD-WEB CHANGES own body tissues (Minagawa and Wada 1984). Kline and Pauly (1998), in the first study of this Ecopath, as shown above, has estimates of trophic type, showed that trophic levels estimated from level as one of its outputs, along with the standard diet compositions, that is, from Ecopath models error of these trophic levels, the square root of the closely correlated with trophic level estimates omnivory index. The many estimates of trophic from stable nitrogen isotopes. levels that emerge from various Ecopath applica- The variance of trophic level estimates can tions helped confirm various generalizations by also be calculated. Given the nature of TL esti- Pimm (1982) and others about the structure of food mates from either stable isotope ratios or diet webs. Also, they allowed going beyond these gen- composition studies, this variance will reflect eralizations. Thus, Pauly and Christensen (1995), feeding at different trophic levels, i.e. omnivory by who had assigned trophic levels to all fish and in- the consumer under study as well as uncertainty vertebrates caught and reported in FAO global fish- concerning the trophic level of its food. Thus, one eries statistics, could show, using between-trophic can define an ‘omnivory index’ (OI) calculated level transfer efficiencies also estimated through from: Ecopath models, that the primary production re- quired to sustain the present world fisheries was n 2 much higher than previously assumed: 8% for the OI =--Â()TLji() TL1 ◊ DCij (10.4) j=1 global ocean and between 25% and 35% for the shelves from which 90% of the world catches orig- where n is the number of groups in the system, TLj inates. Also using time series of the same fisheries is the trophic level of prey j, TL the mean trophic statistics, and trophic levels for the major species, level of the prey (one less than the trophic level Pauly et al. (1998c) demonstrated a steady reduc- of the predator, see above), and DCij is the fraction tion of the mean trophic level of fisheries landings of prey j in the diet of predator i, again as defined from 1950 to the present, suggesting that the above. Rigler (1975) argued that trophic levels fisheries increasingly concentrate on the more were only a ‘concept’, and that mature sciences abundant, small, fast-growing prey fishes and in- should deal with concepts only in the absence of vertebrates near the bottom of aquatic food webs. measurable, actual entities, which alone allow Both of these sets of findings, now validated testing of quantitative hypotheses. The demon- through more detailed local studies quantifying stration that trophic levels estimated from diet human impacts on marine ecosystems (e.g. Pauly composition data and equation (10.3) closely et al. 2000), relied on trophic-level estimates correlate with estimates from stable isotopes of obtained through Ecopath applications, that is, nitrogen not only cross-validates these two diet composition studies that were rendered mutu- methods, but also establishes that trophic levels ally compatible in an ecosystem context. They are not just concepts useful for assigning animals document the utility of the post-Rigler trophic- to various ecological groups, but actual entities, level concept. 220 Chapter 10

10.7 ECOSYSTEM of this theory have been performed so far. One con- STRUCTURE, ODUM’S sisted of forcing an increase in the biomass of top predators in two marine ecosystems, one coastal MATURITY AND and one offshore, and using the Monte Carlo ULANOWICZ’S ‘EcoRanger’ routine to identify parameter values ASCENDANCY which were randomly selected from within the distribution assumed for each of the models’ in- One interesting aspect of food webs is that, once puts and where compatible with these increased constructed, they can be interpreted using various biomasses. This led to increases in all parameters techniques, and the results interpreted in the light related to increased maturity in Table 10.1, no- of various hypotheses on the way food webs should tably detritus recycling (Christensen and Pauly be structured. Some of these hypotheses are dis- 1998). The other test consisted of an application of tinct products, not necessarily connected to other Ecosim, the dynamic version of Ecopath, to be pre- hypotheses. Others are part of broader constructs, sented below. Therein, a short strong increase in such as the theory of Ivlev (1961), or that of Odum the fishing mortality of the small pelagic species (1969). dominant in each system was simulated and the Odum’s theory is interesting in that although it time for the system as a whole to recover was plot- is not strictly quantitative, it makes enough spe- ted. Here again, detritus recycling was the eco- cific qualitative prediction for rigorous tests of system parameter which best correlated with the its validity to be performed (Christensen 1995b). form of resilience implied here (Vasconcellos et al. Thus, food-web models constructed as described 1997). We conclude here that the Ecopath approach above can quantify many of the attributes of can be used to operationalize Odum’s theory and ecosystems that are part of Odum’s theory (Table to test its basic tenets. The theory survived these 10.1). As might be seen, this theory essentially tests. implies that as systems mature, their biomass will The flow charts generated by Ecopath can also tend to increase, especially the biomass of large serve to test Ulanowicz’s theory of ascendancy as a animals with high longevities, and detritus will measure of ecosystem development. This theory increasingly be recycled through a web whose combines the information contents embedded in complexity will tend to increase. Two direct tests the different flows within the system with the

Table 10.1 Selection from Odum’s (1969) list of 24 attributes of ecosystem maturity.

No. in Ecosystem attributes Development stages Mature stages Odum’s list

1 Gross production/respiration >1 or <1 Approaches 1 2 Gross production/biomass High Low 3 Biomass supported/energy flow Low High 4 Net community production High Low 6 Total organic matter Small Large 12 Niche specialization Broad Narrow 13 Size of organisms Small Large 15 Mineral cycles Open Closed 16 Nutrient exchange between organisms and environment Rapid Slow 17 Role of detritus in nutrient recycling Unimportant Important 21 Nutrient conservation Poor Good 22 Stability (resistance to perturbations) Poor Good Ecosystem Models 221 magnitude of these flows to derive a combined prisingly, it turns out, however, that mass-balance measure of information and flow, expressed in models can also be used to generate size spectra of ‘flow bits’(Ulanowicz 1986). Here, results were not the ecosystem, and thus build a bridge between the as unequivocal as in the case of Odum’s theory. hypotheses advanced by marine ecologists and Ascendancy did not correlate positively with those of fisheries scientists. We briefly describe the maturity as expressed using a combination of routine of Ecopath which is used to re-express the parameters derived from Odum’s theory (see biomass of fish, invertebrates and marine mam- Christensen1995b),theimplicationprobably being mals in various ecosystems in the form of stan- that Ulanowicz’s theory is in need of revision. dardized size spectra, whose slope can then be Overall, these two examples illustrate that the compared. This routine, while assuming steady- existing wide availability of quantified food webs state, does the following: constructed using the Ecopath approach can be a • uses the von Bertalanffy growth curves and the boon to theoretical ecology, enabling the testing of values of P/B (i.e. Z) entered for each group in hypotheses that have long remained untested and the model to re-express its biomass in term of a consolidating existing knowledge on the function- size–age distribution; ing of ecosystems. • divides the biomass in each (log) weight class by the time, Dt, required for the organisms to grow out of that class (to obtain the average biomass present 10.8 PARTICLE-SIZE in each size class); SPECTRA IN ECOSYSTEMS • adds the B/Dt values by (log) class, irrespective of the groups to which they belong. Using data from particle-size counters, Sheldon et Figure 10.2 compares size spectra obtained in al. (1972), proposed that the size spectrum of ma- this fashion for a trophic model of the Gulf of rine organisms is a conservative feature of marine Thailand. The spectra are based on 40 ecosystem ecosystems, characterized by a constant slope, for groupings ranging in size from phytoplankton to which they provided the rationale, using thermo- dolphins. The spectra are obtained by running a dynamic considerations. The controversy which model describing the 1973 situation, where fishing ensued, still festering among marine biologists, pressure and resource depletion were moderate, who tend to consider only the small range of organ- with observed fishing pressure through to 1994, ism sizes sampled by automatic plankton particle where fishing pressure was high, and the resources counters, was largely ignored by fisheries scien- severely depleted. Further, a simulation was tists, who quickly established not only that fish performed with fishing pressure removed, and abundances also neatly fit (log) linear size spectra, the long-term equilibrium used to estimate a but that the slope of such spectra reflect the spectrum without fishing. exploitation level to which a multispecies fish The dots on Fig. 10.2 indicate the actual size community is subjected, being steeper where ex- spectrum obtained for the 1973 model. It is far ploitation is high (Pope and Knight 1982; Bianchi from a straight line, but shows a hump in the size et al. 2000). range dominated by benthos. This hump is likely This provides another constraint for multi- to be caused by poor resolution for the benthic species and/or ecosystem models, which should be groupings, and would probably not appear if these expected to generate such spectra as one of their groups were split into finer taxonomic groups than output (as did, the North Sea model of Andersen used for the model presented. and Ursin already described). The three straight lines on Fig. 10.2 show that Individual-based models such as OSMOSE the slope of the size spectra, as expected, increases (Shin and Cury 1999) can generate such spectra, as with fishing pressure, here from (-0.17) for the may perhaps be expected (Shin 2000; see also Huse unfished situation, to (-0.18) for the moderately et al., Chapter 11, this volume). Perhaps more sur- fished, and to (-0.22) for the situation with high 222 Chapter 10

No fishing (–0.17) 2.0 Moderate fishing (–0.18) 1.5 High fishing (–0.22) 1.0 )

–2 0.5 0.0 –0.5 –1.0

Biomass (log g m –1.5 –2.0 –2.5 –0.6 0.3 1.8 3.3 3.8 Body weight (log g)

Fig. 10.2 Ecosystem-level size spectra for the Gulf of Thailand. The spectra are derived based on a trophic model with 40 ecosystem groupings (see www.ecopath.org). The thin line associated with ‘moderate fishing’ describes the 1973 situation in the Gulf, while the broken line indicates the slope of the model situation at long-term equilibrium without fishing, and the thick line indicates the 1994 situation, with resources severely depleted by over-fishing. Note that the slopes of the spectra (in brackets) increase with fishing pressure.

fishing pressure. The increased slope is caused by ddBtgi=◊ iÂÂ Q ij - Q ji +- I i() M0 i ++ FEB i i i , the removal of virtually all higher trophic-level (10.5) organisms in the Gulf during the time span studied (see Christensen 1998; Pauly et al. 1998b). where dB/dt is the rate of biomass change, g the The example given here only serves as a taster growth efficiency (i.e. P/Q), F the fishing mortality for how size spectra may be of use as ecosystem (i.e. Y/B), M0 the natural mortality (i.e. excluding indicators of exploitation level, and no gener- predation), I is the immigration rate, E the emigra- alizations have been drawn so far of how the slopes tion rate, and Qij (Qji) the consumption of type j (i) of the spectra relate to exploitation. We anticipate, biomass by type i (j) organisms. however, that the ease with which ecosystem- This set of coupled differential equations could level size spectra can now be constructed through be easily integrated as they are over time, thus the Ecopath approach will lead to a blossoming yielding a simulation model with the help of of spectral analysis in the foreseeable future, which various scenarios resulting from changes and that valuable insights will be gained in the in fishing mortality F, could be explored. This process. simulation model, however, would be of the Lotka–Volterra type, wherein the amount con- sumed of a given prey ‘i’ is proportional to the 10.9 ECOSIM, REFUGIA product of its biomass times the biomass of its AND TOP-DOWN VS. predator(s). Such ‘top-down controlled’ systems, BOTTOM-UP CONTROL however, are inherently unstable, and usually fluc- tuate in unrealistic fashion. As animals do not live The system of coupled linear equations that is in reaction vats, modelling predation must reflect behind a balanced Ecopath model (see equation the ability of potential prey to hide or camouflage 10.1) above can be re-expressed in terms of the themselves, or generally to evolve strategies that change implied by: limit their exposure to predators (see Krause et al., Ecosystem Models 223

Chapter 13, Volume 1). In Ecosim, the dynamic terms of the implementation presented here, inter- counterpart to Ecopath, the existence of physical mediate control, which is a form of control that is or behavioural refugia is represented by prey bio- neither fully top-down, nor bottom-up, is required mass consisting of two elements: one, vulnerable for simulated ecosystems to behave in realistic to predators, the other invulnerable. It is then the fashion, this being a major finding obtained rate of transfer between these two partial compo- through Ecosim (Walters et al. 1997, 1999; Pauly et nents of a prey’s biomass which determine how al. 2000; Christensen et al. 2000; and www.eco- much of the prey can be consumed by a predator. path.org may be consulted for further information When the exchange rate is high, part of the bio- on this rapidly evolving software). mass which is vulnerable is quickly replenished and hence we still have top-down control and Lotka–Volterra dynamics. On the other hand, 10.10 SPATIAL when the replenishment of vulnerable biomass CONSIDERATIONS IN pool is set to be slow, it is essentially that slow rate ECOSYSTEM MODELLING which determines how much the predators can consume. We speak here of ‘bottom-up control’, As mentioned in the introduction, adding com- since it is the dynamics of the prey which shape the plexity to ecosystem models does not necessarily ecosystem. Issues of top-down versus bottom-up make them ‘better’. Rather, to increase the useful- control are addressed further in the food webs ness of models, what is required is to identify those chapter by Polunin and Pinnegar (Chapter 14, improvements of existing models for which the Volume 1) and Persson (Chapter 15, Volume 1). gain in new insights outweighs the added com- These different control types can be represented plexity and data requirements. As it turned out, by replacing Qij in the above equation by: the major improvements that can be added to models such as Ecopath and Ecosim are spatial

QvaBBvvij=+¢+ ij ij i j( ij ij aB ij j ), (10.6) considerations, capable of representing explicitly some of the refugia implied in the above Ecosim where vij and vij¢ represent rates of behavioural ex- formulation. change between vulnerable and invulnerable state The formulation developed for this, still based and aij represents the rate of effective search by on Ecopath parameterization and its inherent predator j for prey i, i.e. the Lotka–Volterra mass- mass-balance assumptions, is one wherein the action term. ecosystem is represented by say, a 20 ¥ 20 grid of Thus, Ecosim, which allows users to change the cells with different suitability to the different rates of exchange, allows testing the effect of as- functional groups in the system. Movement rates sumptions about bottom-up vs. top-down control. are assumed symmetrical in all directions around As it turns out, these effects are profound: pure the cell, but are higher in unsuitable habitat. Their top-down control, which is the Lotka–Volterra exact value is not important. The survival of assumption, generates, upon the smallest shock, various groups and their food consumption are violent oscillations such as do not occur in real assumed higher in suitable habitat, but they other- ecosystems. Conversely, under bottom-up control, wise consume prey as they do in Ecopath and depletion of one species, as, for example through Ecosim, as they encounter them within a given fishing, tends to affect the biomass of only that cell. species and less strongly its key prey and predators; Starting from the Ecopath baseline where func- the system as a whole generally remains unaffect- tional groups are distributed evenly over suitable ed even when the species in question is one of habitats, Ecospace simulation iterates towards a its major components (the astute reader will note solution wherein the biomass of all functional this to be the unstated assumption behind single- groups is spread over a number of cells, given the species population dynamics). Thus, at least in predation they experience and the density of prey 224 Chapter 10 organisms they encounter in each cell. The distrib- tions, the sequence of cells which, when added to ution maps thus predicted can be compared with that initial seed, will contribute most to overall existing distribution maps and inferences drawn benefits, either in terms of market values calcul- about one’s understanding of the ecosystem in ated as species caught multiplied by their price, question and the functional group they are in. The minus cost of catching them, and/or in terms of rich patterns obtained by the application of this their existence values, for example, as whale abun- approach to a number of Ecopath files, thus turned dance to the whale-watching industry. Results of into spatial models, suggest that the broad pattern various Ecoseed runs for different systems show, of the distribution of aquatic organisms can be in accordance with theoretical expectation, that straightforwardly simulated (Walters et al. 1998). MPAs should be large relative to exploited areas. One particularly interesting aspect of Ecospace Important here is that MPAs simulated this way is that it allows for explicit consideration of eco- consider all trophic interactions between the system effects when evaluating the potential im- species included in an ecosystem model and not pact of Marine Protected Areas (MPA) in a given only the expected biomass increase of a few charis- ecosystem, thus allowing for a transition towards matic species. ecosystem-based fisheries management, our last 2 An optimization approach structured around topic. Ecosim’s ability to identify the mix of fleet- specific fishing mortality which optimizes cumu- lative benefits over a set period, e.g. 30 years, under 10.11 TOWARDS A any of the following constraints: (a) maximizing net return to the fishery, which TRANSITION FROM SINGLE- generally involves moderate mortality on valu- SPECIES TO ECOSYSTEM- able target species and a small overall level of BASED MANAGEMENT effort; (b) maximizing employment, which means iden- The requirement for ecosystem-based manage- tifying the fleet configuration which sustainably ment derives its validity from a stark set of alterna- exploits the ecosystem, albeit with an effort level tives: either the beginning of the 21st century will as high as possible. This high effort leads to high see a transition towards ecosystem-based man- employment; agement in most areas currently exploited by (c) maximizing ecosystem maturity, by identify- major fisheries, or it will see the destruction of the ing the fleet configuration which maximizes, for ecosystems exploited by these same fisheries. all groups in a system, the sum of product of bio- Until recently, the call for ecosystem-based mass and B/P. The latter is in accordance with management could be dismissed by regulatory Odum (1969) who predicts the highest product of agencies intent on continuing business as usual B and B/P for the long-lived organisms typical of because ecosystem management tools were not mature systems; available. The sets of conceptual tools mentioned (d) mandated rebuilding, wherein the fleet config- above, notably Ecopath, Ecosim and Ecospace can, uration is sought which enables faster rebuilding however, be used to identify the key elements of towards a set threshold as required. This could management strategies that would enable fish- result from a consequence of a legal decision; eries to be sustained by sustaining the ecosystem (e) optimizing a mix of (a)–(d) wherein any of these in which they are embedded. The two major tools can be given different weights. available for this are: These two approaches combined and used inde- 1 Ecoseed: a routine for seeding marine protected pendently allow scientists, for any ecosystem for areas (MPAs), originally covering only one cell which a suitably detailed Ecopath model has been within the spatial grid in an Ecopath/Ecospace previously constructed, to explore the implica- map, then identifying, by brute-force computa- tions of various policies, and to quantify their out- Ecosystem Models 225 comes in economic and social terms, and in the form of the ecosystem biodiversity they imply. A REFERENCES very important aspect of these simulations is that they invite a discussion of how we want ecosys- Allen, K.R. (1971) Relation between production and biomass. Journal of the Fisheries Research Board of tems to look in the future, and the discussion takes Canada 28, 1573–81. place in a quantifiable manner. Alward, G.L. (1932) The Sea Fisheries of Great Britain and Ireland. Grimsby: Albert Gait. Andersen, K.P. and Ursin, E. (1977) A multispecies exten- 10.12 CONCLUSIONS sion to the Beverton and Holt theory of fishing, with accounts of phosphorus circulation and primary The concepts and tools for ecosystem modelling production. Meddelelser fra Danmarks Fiskeri og Havundersogelser, N.S. 7, 319–435. presented above allow for a transition towards Beverton, R.J.H. and Holt, S.J. (1957) On the dynamics of ecosystem-based management of fisheries. Obvi- exploited fish populations. United Kingdom Ministry ously, uncertainty remains a major issue in such of Agriculture, Fisheries and Food. Fishery Investiga- analyses. For example, the science of how whole tions Series II, vol. xix, 533 p. ecosystems respond to changes in management re- Bianchi, G., Gislason, H., Graham, K., Hill, L., Jin, X., mains very weak (Polunin and Pinnegar, Chapter Koranteng, K., Manickchand-Heileman, S., Payá, I., Sainsbury, K., Arreguin-Sánchez, F. and Zwanenburg, 14, Volume 1). However, not using tools such as de- K. (2000) Impact of fishing on size composition and scribed here will not reduce uncertainty. The way diversity of demersal fish communities. ICES Journal to reduce uncertainty about ecosystem function of Marine Science 57, 558–71. and the impacts of fishing thereon is to construct Christensen, V. (1995a) A multispecies virtual popula- representations of these ecosystems and to probe tion analysis incorporating information of size and age. their behaviour by posing intelligent questions ICES CM 1995/D: 8. about the process, and intelligently interpret- Christensen, V. (1995b) Ecosystem maturity – toward quantification. Ecological Modelling 77, 3–32. ing the answers. The effect of uncertainties will Christensen, V. (1998) Fishery-induced changes in manifest themselves and they will then have marine ecosystems: insights from the Gulf of Thai- to be confronted using whatever new tools might land. Journal of Fish Biology 53 (Suppl. A), 128–42. become available. On the other hand, the effect of Christensen, V. and Pauly, D. (1992) The ECOPATH II – a fishing on ecosystems is not uncertain. The time software for balancing steady-state ecosystem models has now come for us to generalize from the accu- and calculating network characteristics. Ecological mulated experience of a century of marine and Modelling 61, 169–85. Christensen, V. and Pauly, D. (eds) (1993) Trophic models fisheries research; the tools presented above pro- of aquatic ecosystems. ICLARM Conference Proceed- vide a context wherein these generalizations can ings No. 26. International Center for Living Aquatic be made. Resources Management, Manila. Christensen, V. and Pauly, D. (1998) Changes in models of ecosystems approaching carrying capacity. Ecologi- ACKNOWLEDGEMENTS cal Applications 8 (1), 104–9. Christensen, V., Walters, C.J. and Pauly, D. (2000) Ecopath with Ecosim: a User’s Guide. Fisheries We thank Carl Walters for his collaboration, and Centre, University of British Columbia, Vancouver: for crucial contributions towards the development and Penang: International Center for Living Aquatic of Ecosim and Ecospace, both of which make Resources Management. Ecopath more useful than it had ever been. We Cortés, E. (1999) Standardized diet compositions and also thank the hundreds of Ecopath users; without trophic levels of sharks. ICES Journal of Marine them, the generalizations presented here would Science 56, 707–17. Cotte, M.J. (1944) Poissons et animaux aquatiques aux never have emerged. Finally, thanks to Deng temps de Pline: Commentaires sur le Livre IX de ‘His- Palomares, without whom this contribution toire Naturelle de Pline. Paris: Paul Lechevalier. would never have been written. Cousins, S.H. (1985) The trophic continuum in marine 226 Chapter 10

ecosystems: structure and equations for a predictive Lindeman, R.L. (1942) The trophic dynamic concept in model. In: R.E. Ulanowicz and T. Platt (eds), Ecosystem ecology. Ecology, 23 (4), 399–418. theory for biological oceanography, Canadian Bulletin Lotka, A.J. (1925) Elements of Mathematical Biology. of Fisheries and Aquatic Science 213, 76–93. New York: Dover Publications. Darwin, Charles (1859) The Origin of Species. London: MacDonald, J.S. and Green, P.H. (1983) Redundancy of John Murray. variables used to describe importance of prey species in Froese, R. and Pauly, D. (2000) FishBase 2000: Concepts, fish diet. Canadian Journal of Fisheries and Aquatic Design and Data Sources. Los Baños, Philippines: Sciences 40, 635–7. ICLARM. MacKay, A. (1981) The generalized inverse. Practical Gause, G.F. (1934) The Struggle for Existence. Baltimore, Computing (September), 108–10. Md.: Williams and Wilkins. Mathisen, O.A. and Sands, N.J. (1999) Ecosystem model- Gayanilo, F.C., Jr., Sparre, P. and Pauly, D. (1996) The ing of Becharof Lake, a Sockeye salmon nursery lake FAO-ICLARM Stock Assessment Tools (FiSAT) User’s in Southwestern Alaska. pp. 685–703 In: Ecosystem Guide. FAO Computerized Information Series/ Approaches for Fisheries Management. Alaska Sea Fisheries 8. Grant College Program AK-SG-99-01. Golley, F.B. (1993) A History of the Ecosystem Concept in Minagawa, M. and Wada, E. (1984) Stepwise enrichment Ecology: More than the Sum of its Parts. New Haven, of 15N along food chains: further evidence and the rela- Ct.: Yale University Press. tion between 15N and animal age. Geochimica and Grigg, R.W. (1982) Darwin Point: a threshold for atoll Cosmochimic Acta 48, 1135–40. formation. Coral Reefs 1, 29–34. Munro, J.L. (1983) Assessment of the potential pro- Ivlev, V.S. (1961) Experimental ecology of the feeding of ductivity of Jamaican waters. In: J.L. Munro (ed.) fishes. New Haven, Ct.: Yale University Press. Caribbean Coral Reef Fisheries Resources. ICLARM Jones, R. (1982) Ecosystems, food chains and fish yields. Studies and Reviews. 7, pp. 232–48. International In: D. Pauly and G.I. Murphy (eds) Theory and Manage- Center for Living Aquatic Resources Management, ment of Tropical Fisheries. ICLARM Conference Manila. Proceedings, 9, pp. 195–239. International Center for Murawski, S.A., Lange, A.M. and Idoine, J.S. (1991) An Living Aquatic Resources Management, Manila. analysis of technological interactions among Gulf of Kirkwood, G.P. (1982) Simple models for multispecies Maine mixed-species fisheries. pp. 237–52. In: N. Daan fisheries, In: D. Pauly and G.I. Murphy (eds). Theory and M.P. Sissenwine (eds) Multispecies Models and Management of Tropical Fisheries. ICLARM Relevant to Management of Living Resources. ICES Conference Proceedings 9, pp. 83–98. International Marine Science Symposia (193), pp. 237–52. Center for Living Aquatic Resources Management, NRC (1999) Sustaining Marine Fisheries. National Manila. Research Council. Washington DC: National Acade- Kline, T.C. Jr. and Pauly, D. (1998) Cross-validation of my Press. trophic level estimates from a mass-balance model of Odum, E.P. (1969) The strategy of ecosystem develop- Prince William Sound using 15N/14N data. pp. ment. Science 164, 262–70. 693–702 In: T.J. Quinn II, F. Funk, J. Heifetz, J.N. Odum, W.E. and Heald, E.J. (1975) The detritus-based Ianelli, J.E. Powers, J.F. Schweigert, P.J. Sullivan and C.- food web of an estuarine mangrove community. In: L.E. I. Zhang (eds). Proceedings of the International Sym- Cronin (ed.) Estuarine Research. Vol. 1. New York: posium on Fishery Stock Assessment Models. Alaska Academic Press, pp. 265–86. Sea Grant College Program Report No. 98-01. Alaska Opitz, S. (1993) A quantitative model of the trophic Sea Grant, Fairbanks. interactions in a Caribbean coral reef ecosystem. In: Koslow, J.A. (1997) Seamounts and the ecology of deep V. Christensen and D. Pauly (eds) Trophic Models of sea fisheries. American Scientists 85 (2), 168–76. Aquatic Ecosystems. ICLARM Conference Proceed- Laevastu, T. and Larkins, H. (1981) Marine Fisheries ings 26, pp. 259–67. Ecosystem: Its Quantitative Evaluation and Manage- Pauly, D. and Christensen, V. (1995) Primary production ment. Farnham, Surrey: Fishing News Books. required to sustain global fisheries. Nature (374), Larkin, P.A. and Gazey, W. (1982) Application of ecologi- 255–7. cal simulation models to management of tropical Pauly, D., Soriano-Bartz, M.L. and Palomares, M.L.D. multispecies fisheries. In: D. Pauly and G.I. Murphy (1993) Improved construction, parametrization and (eds) Theory and Management of Tropical Fisheries. interpretation of steady-state ecosystem models. In: ICLARM Conference Proceedings 9, pp. 123–40. V. Christensen and D. Pauly (eds) Trophic Models of International Center for Living Aquatic Resources Aquatic Ecosystems. ICLARM Conference Proceed- Management, Manila. ings, 26, pp. 1–13. Ecosystem Models 227

Pauly, D., Trites, A., Capuli, E. and Christensen, V. Thompson, D.W. (translator) (1910) Historia Anima- (1998a) Diet composition and trophic levels of marine lium, Vol. 4. In: J.A. Smith and W.D. Ross (eds) The mammals. ICES Journal of Marine Science 55, 467–81. Work of Aristotle. Oxford: Clarendon Press. Pauly, D., Froese, R. and Christensen, V. (1998b) How per- Tort, P. (ed.) (1996) Dictionnaire du Darwinisme et de vasive is ‘fishing down marine food webs’: response. L’Evolution. Paris: Presses Universitaires de France. Science 282 (5393), 1383a, 20 November. Trites, A.W., Livingston, P.A., Mackinson, S., Pauly, D., Christensen, V., Dalsgaard, J., Froese, R. and Vasconcellos, M.C., Springer, A.M. and Pauly, D. Torres Jr, F.C. (1998c) Fishing down marine food webs. (1999) Ecosystem changes in and decline of marine Science 279, 860–3. mammals in the Eastern Bering Sea: testing the ecosys- Pauly, D., Christensen, V. and Walters, C. (2000) Ecopath, tem shift and commercial whaling hypothesis. Fish- Ecosim and Ecospace as tools for evaluating ecosystem eries Centre Research Reports, 7(1). impact of fisheries. ICES Journal of Marine Science 57, Ulanowicz, R.E. (1986) Growth and Development: 697–706. Ecosystem Phenomenology. New York: Springer Pimm, S.L. (1982) Food Webs. Population and Commu- Verlag. nity Series. London: Chapman & Hall. Ursin, K. (1973) On the prey preference of cod and dab. Polovina, J.J. (1984) Model of a coral reef ecosystem I. The Meddelelser fra Danmarks Fiskeri og Havunderso- ECOPATH model and its application to French Frigate gelser N.S. 7, 85–98. Shoals. Coral Reefs 3, 1–11. Vasconcellos, M., Mackinson, S., Sloman, K. and Pauly, Polovina, J.J. and Tagami, D.T. (1980) Preliminary results D. (1997) The stability of trophic mass-balance models from ecosystem modeling at French Frigate Shoals. of marine ecosystems: a comparative analysis. Ecologi- In: R.W. Griggs and R.T. Pfund (eds) Proceedings of cal Modelling 100, 125–34. the Symposium on the Status of Resources – Investiga- Volterra, V. (1926) Variations and fluctuations of in- tions in the Northwestern Hawaiian Islands, 24–25 dividuals of animals living together. pp. 409–48 April (1980) University of Hawaii, Sea Grant Misc. In: R.N. Chapman (ed.) (1931) Animal Ecology, Reports UNIHI-SEAGRANT-MR-80-04, pp. 286–98. with Special Reference to Insects. New York: Mc- Pope, J.G. and Knight, B.J. (1982) Multispecies Graw-Hill. approaches to fisheries management. In: M.C. Mercer Walters, C.J. and Juanes, F. (1993) Recruitment limita- (ed.) Canadian Special Publication of Fisheries and tion as a consequence of natural selection for use of Aquatic Science, no. 59, pp. 116–18. restricted feeding habitats and predation risk taking Rigler, F.H. (1975) The concept of energy flow and nutri- by juvenile fishes. Canadian Journal of Fisheries and ent flow between trophic levels. In: W.H. van Dobben Aquatic Sciences 50, 2058–70. and R.H. Lowe-McConnel (eds). Unifiying Concepts Walters, C., Christensen, V. and Pauly, D. (1997) Struc- in Ecology. The Hague: W. Junk B.V. Publishers, pp. turing dynamic models of exploited ecosystems from 15–26. trophic mass-balance assessments. Reviews in Fish Sheldon, R.W., Prackash, A. and Sutclifffe, W.H. (1972) Biology and Fisheries 7 (2), 139–72. The size distribution of particles in the ocean. Limn. Walters, C., Pauly, D. and Christensen, V. (1998) Eco- Oceanogr. 17, 327–40. space: prediction of mesoscale spatial patterns in Shin, Y. (2000) Interaction trophiques et dynamiques des trophic relationships of exploited ecosystems, with populations dans les écosystèmes marins exploités: emphasis on the impacts of marine protected areas. approche par modélisation-individus-centrée. Thesis, Ecosystems 2, 539–54. Université Paris 7, 245 pp. Walters, C.J., Pauly, D., Christensen, V. and Kitchell, J. Shin, Y. and Cury, P. (1999) OSMOSE: a multispecies (1999) Representing density dependent consequences individual-based model to explore the functional role of life history strategies in aquatic ecosystems: Ecosim of biodiversity in marine ecosystems. In: Ecosystem II. Ecosystems 3, 70–83. Approaches for Fisheries Management. Alaska Sea Wigner, E. (1960) The unreasonable effectiveness Grant College Program AK-SG-99-01, pp. 593–608. of mathematics in the natural sciences. Com- Thienemann, A.F. (1925) Der See als Lebenseinheit. munications on Pure and Applied Mathematics 13, Naturwissenschaften 13, 589–600. 1–13. 11 Individual-based Models

GEIR HUSE, JARL GISKE AND ANNE GRO VEA SALVANES

11.1 INTRODUCTION al. 1988; Caswell and John 1992). Even though the structured models do partition a population, it is Variability in states and traits is pronounced at difficult to incorporate features such as spatial de- all life stages of fish populations. This is mainly tail into these models, because it is rarely reason- caused by spatial and temporal fluctuations in the able to assume that all individuals of a certain state environment, which in turn result in variation in occupy the same position in space. Structured feeding, growth history and adaptation among models also generally divide the population by one individuals in a population (Magurran 1986; variable, but real individuals may differ with Salvanes and Hart 2000). This individual vari- regard to many variables. ability often has consequences for population These features can be implemented in abundance, spatial distribution and fecundity. Tra- individual-based models (IBM), which keep track ditional ecological models, such as the Lotka– of each individual in a population (DeAngelis et al. Volterra competition models (Pauly and Chris- 1979; L¢ omnicki 1988; Huston et al. 1988). In these tensen, Chapter 10, this volume) or the Ricker models individuals can be characterized by state model for stock recruitment (Myers, Chapter 6, variables such as weight, age and length, and they Volume 1), do not take into account variability also allow behavioural strategies to be imple- among individuals. Rather, these models use mented in a spatial context. This allows the prop- population abundance as a single characteristic erties of a population to be described by the that defines the population dynamics and it is as- properties of its constituent individuals. Model sumed that individuals are identical. Consequent- validations against data can be done at the individ- ly, features such as size structure, known to be ual level, which is an appealing property because important in fish population dynamics, are left out observations often are performed on single indi- from the model specifications. viduals. Also, models based on individuals benefit The traditional models have been developed from having the same basic unit as natural selec- into structured population models where the tion (Darwin 1859; Williams 1966). These issues population is divided according to age, stage or make individual-based modelling an appealing some physiological criterion (Metz and Diekman tool in ecology. The approach is not new per se, but 1986; Sparre and Hart, Chapter 13, this volume). rather it is analogous to the old reductionism that Such structured models have proven successful has been very successful in empirical sciences for many applications in ecology and fisheries sci- (L¢ omnicki 1988). ence, and the use of differential equations or ma- The early work of DeAngelis et al. (1979) and trix models allows analytical solutions (Huston et Beyer and Laurence (1980) on modelling growth Individual-based Models 229 and survival of largemouth bass (Micropterus salmoides) and winter flounder (Pseudopleu- 11.2 SPECIFYING ronectes americanus) respectively, set the scene INDIVIDUALS IN IBMs for extensive later use of IBMs in early life history 11.2.1 The attribute vector studies. The major motivation for individual- based modelling of these early life stages has been Here we refer to IBMs as models that treat individ- to explore causes of recruitment variability to uals as explicit entities, the so-called i-state con- commercial fish stocks, an issue that has prevailed figuration models (Caswell and John 1992). We and been studied empirically in fishery science will focus mostly on these models, but we will also since the work of Hjort (1914; see also Myers, discuss structured models that sometimes have Chapter 6, Volume 1 and Smith, Chapter 4, this been classified as IBMs (Caswell and John 1992). volume). In order to simulate the survival and spa- The i-state refers to individual features such as tial distribution of early life stages of fish cohorts, body weight, energy reserves and sex, while corre- it is important to take account of individual vari- sponding p-states represent the whole population ability, since the eventual survivors tend to differ such as population abundance and average i-states from average individuals at earlier stages (Crowder of the population. It can be fruitful to illustrate the et al. 1992). Studies of early life history in fish have concept of IBMs by using an attribute vector Ai consequently been one of the topics where IBMs (Chambers 1993), which contains all the states have been applied most extensively (Grimm 1999). ami used to specify an individual i such as age, Although the individual-based modelling ap- weight, sex, hormone levels and spatial coordi- proach was initiated in the late 1970s, it is only nates (xi,yi,zi) at time t: since the influential review of Huston et al. (1988) that it has been applied extensively in ecology. Still A i = ()aa123iii, , a ,... amxyzt iiii , , , , . (11.1) it has not been established whether or not IBMs The greater the attribute vector, the more differ- give fundamentally different answers from the ences between individuals can be specified within classical ecological models because very few the model. paradigmatic studies have challenged the classic In structured models (Metz and Diekman 1986; models (Grimm 1999). Nevertheless, IBMs pro- Tuljapurkar and Caswell 1997), populations are vide a flexible tool for simulating individuals and divided into stages based on some key variable, for populations. example age, which is commonly applied in, for In contrast to earlier reviews of IBMs (Huston et example, the virtual population analysis of quanti- al. 1988; DeAngelis and Gross 1992; Grimm 1999), tative fisheries science. Using the attribute vector this chapter focuses on the use of individual-based concept, one may describe structured models as: modelling in fish ecology and fisheries science. Rather than providing a balanced review of what A j = ()snjj,, (11.2) has been done using IBMs, we will present topics that modellers of fish populations might need to where sj is stage j and nj is the number of individ- deal with and then provide some relevant exam- uals of the population in stage j. The changes in Aj ples for the particular fields. As a result of this we can then be projected using models such as Leslie will not cover many replicate studies dealing with matrix models or partial differential equations, similar issues, but instead present a wider range of which can be called i-state distribution models applications taking advantage of the individual- (Caswell and John 1992). In IBMs, each individual based approach. We start out with a presentation is specified independently, which means that the of the IBM concept, including development and number term nj of equation (11.2) is 1, and essen- evaluation, and provide some recipes for making tially removed from the attribute vector. However, different kinds of IBMs. Then we move on to a re- even though the individual-based structure is view of existing literature on IBMs. appealing, it is virtually impossible to simulate 230 Chapter 11 even small fish stocks on a truly individual basis is more flexible than the structured models because of the great abundances involved. To (DeAngelis and Rose 1992), and allows a simple allow the advantages of the individual-based ap- way of scaling IBMs to realistic fish stock abun- proach and still be able to simulate large popula- dances. Structured models and configuration mod- tions such as fish stocks, the super-individual els may differ in their applicability depending on approach was introduced (Scheffer et al. 1995). A the topic in question, but in cases where both ap- super-individual represents many identical indi- proaches can be applied they tend to give similar viduals and in this case the number of such identi- predictions (DeAngelis et al. 1993). cal siblings (ns) thus becomes an attribute of the super-individual: 11.2.2 The strategy vector

As = ()aa123s, s , a s ,... amxyznt sssss , , , , , . (11.3) In addition to possessing states, real individuals have adaptive traits, such as life history and behav- where As is the attribute vector of super-individual ioural strategies that specify how they should live s. Mortality operates on the super-individual and their life. The previous lack of IBM studies involv- the number of siblings of each super individual is ing life-history strategies and behaviour of individ- thus decreased in proportion to the mortality rate uals could be due in part to a lack of appropriate (Scheffer et al. 1995). This is an efficient way of techniques for implementing these features. How- maintaining the individual-based structure, and ever, adaptive traits can be modelled by introduc- still be able to simulate the large population sizes ing a strategy vector Si that specifies the adaptive that occur in natural populations. When the ns gets traits, such as life-history traits or behaviour, of an below a threshold value, the way the mortality rate individual: operates can be changed to probabilistic mortality using Monte Carlo techniques (see below) for the Si = ()bb123iii, , b ,..., bm i , (11.4) remaining siblings. Populations with high mortal- ity, such as fish populations, can effectively be where bmi is the adaptive trait m of individual i. simulated by replacing a dead super-individual The strategy vector may be considered as analo- through random resampling from the live portion gous to a biological chromosome as in the genetic of the population (Rose et al. 1993). The internal algorithm (Holland 1975), but Si may also be up- number ns of the donor individual is then divided dated during the individual’s life as a way to simu- in two and the dead super-individual inherits late learning. For example, this can be done using the attributes of the donor. Thus one may keep reinforcement learning (e.g. Ackley and Littman the number of super-individuals constant while 1992) by allowing rewards for advantageous behav- changing the number of individuals that actually iours and punishments for unprofitable ones. This are represented by each super-individual. process allows the individual to produce increas- The aggregation in super-individuals has ingly more favourable behaviours as it learns about some obvious similarities with the structured its environment. The combination of attribute models (Metz and Diekman 1986; Tuljapurkar and vectors and strategy vectors thus enables most Caswell 1997). However, the structured popu- relevant characteristics of individuals to be imple- lation models are based on partial differential mented in IBMs. equations or Leslie matrices, while the super- individual approach still maintains the same structure and representation of processes as in the 11.3 FEATURES OF IBMs discussed above. In structured models all INDIVIDUAL-BASED MODELS stages are usually assumed to experience the same environment and therefore to respond similarly IBMs can be classified by the degree to which (Caswell 1996). The super-individual approach different factors such as bookkeeping (i.e. the Individual-based Models 231

Bookkeeping process (Juanes et al., Chapter 12, Volume 1), Growth and bioenergetics (Jobling, Chapter 5, Volume 1). Survival Mechanistic models are used extensively in IBMs to represent individual processes such as percep- tion, predator and prey encounters, and bioener- Local interactions Space getics. These are indicated by bold squares in Fig. Links between IBM Active movement 11.2, which shows a flow chart for events related individuals Drift to feeding, growth and predation of fish larvae. In addition to these individual processes, the envi- ronment is also often represented through mecha- Adaptive traits nistic models of features such as light and ocean Behaviour circulation. For a thorough discussion of mecha- Life history strategies nistic models in fish biology see Giske et al. (1998) and Carlotti et al. (2000). Fig. 11.1 Schematic classification of various types of IBMs. Most IBMs keep track of growth and survival of individuals in a population. Implementation of space 11.3.2 Bookkeeping allows a wider range of studies to be undertaken. A further increase in the complexity and flexibility of A central aspect of IBMs is the bookkeeping of indi- models to incorporate adaptive traits such as behaviour viduals. As shown above, this is facilitated using and life history strategies may further promote our the attribute vector. The entire population is understanding of how individuals relate to each other. tracked using an attribute matrix with dimen- The simulation of local interactions involves a high sions equal to the number of attributes times resolution both in spatial detail and behavioural actions. the number of individuals in the population. All events that may occur in a period of a time step will be addressed sequentially in the IBM, as if they continuous update of the attribute vector), space, appeared one by one, and it may be important to adaptive traits and local interactions are specified analyse the processes in a particular sequence. in the model (Fig. 11.1). Depending on the nature While an organism may be eaten after it has and scale of the problem of interest one may devel- starved to death, it may not starve after having op a simple bookkeeping model, or complex been eaten. The conceptual framework for book- models taking into account more aspects of indi- keeping illustrated in Fig. 11.2 is characteristic of viduals. All IBMs contain a bookkeeping proce- most IBMs, which deal with growth and survival of dure, but they need not contain the other features fish. Monte Carlo simulations are continuously listed in Fig. 11.1. Another common feature used to decide the outcome of feeding and mor- of IBMs is to represent ecological processes by tality processes. mechanistic models. 11.3.3 Monte Carlo simulations 11.3.1 Mechanistic models Monte Carlo simulations are often applied in As opposed to empirically fitted models, mecha- IBMs, but this approach is rarely well defined. In nistic models aim at representing the actual general, Monte Carlo techniques involve draw- process that is taking place in more detail. Thus in- ing random numbers from some probability dis- stead of simply fitting growth rate as an empirical tribution, in a way similar to gambling situations relationship of, for example, size, a mechanistic from which the name derives. The usual way model of growth will address the various processes to apply this concept in IBMs is to assume a pro- involved. These include encounters with prey bability for some event and then draw a number (Mittelbach, Chapter 11, Volume 1), the ingestion from a random-number generator to determine 232 Chapter 11

Get next individual

Prey Calculate prey availability encounter rate and feeding

no yes Record Feed? Starve? starvation mortality yes no Calculate growth and new larval size

Calculate encounter Predator rate with predators Fig. 11.2 A conceptual model field and probability of of a sequential representation being captured of events related to growth and survival of fish larvae. The same sequence of events must be yes Record repeated for every individual in the Eaten? predation population, before starting again mortality with the first individual in the no next time step. Bold squares indicate where mechanistic Add to models are applied and ovals individuals surviving indicate where bookkeeping to next time step is performed. (Modified from Crowder et al. 1992.)

the outcome (Judson 1994). For example, one 11.3.4 Spatial detail might imagine a certain probability that an individual dies during a time step. A random Traditional ecological models specify space as a number is then drawn, and if the random number homogenous well-mixed compartment (Fig. 11.3). is smaller than the mortality risk, the individual In metapopulation models, space is implemented dies. Given that the random numbers are uni- as habitable patches surrounded by non-habitable formly distributed, the probability of drawing a areas (Levins 1969). Metapopulation models have number within an interval is equal to the size of been applied to some problems in fisheries sci- the interval. If the mortality risk is 0.1, there is a ence, for example to explain population struc- 10% probability that the random number will be turing in herring stocks (McQuinn 1997). A still between 0 and 0.1, and a similar probability of finer spatial resolution is to arrange space in dying. This can be carried out to determine two- or three-dimensional grids where essentially whether an individual survives, eats food or ob- any spatial scale can be represented (Fig. 11.3). tains mating and so forth. IBMs can be used in all these spatial arrangements. Individual-based Models 233

Single Metapopulation Grid-based Table 11.1 Pseudo-code for implementing spatial detail compartment models in studies of larval fish.

1 Create current vector from a hydrodynamic model. 2 Add component of active individual movement. 3 Integrate velocity vector over the duration of the time step to produce individual trajectories. 4 Find temperature, food concentration, light intensity, predator Fig. 11.3 Different ways of representing spatial detail density, etc. along trajectory. in models. The complexity in spatial representation 5 Estimate food encounter rate, predation risk, bioenergetics etc. increases in the direction of the arrows. 6 Find state (spatial position, physiological state, alive/starved/eaten) of organism at the end of time step.

The development of IBMs has made it possible to implement space in ecological models in a more re- Werner et al. 1996; Lynch et al. 1998; Asplin et al. alistic manner than structured models allow. Grid 1999). The drift path may also be impacted by models have been extensively applied in fisheries swimming of the larvae, which is then added to the science for studies of the distribution of eggs and physical vector (Bartsch et al. 1989; Bartsch and larvae (Bartsch et al. 1989; Werner et al. 1996). An Knust 1994). Once the drift path of a larva is found, example of how such models are set up is provided the temperature, food, light intensity etc. along below. this path can also be estimated (Table 11.1). A more thorough description of this coupling is given by Carlotti et al. (2000). Recipe for simulating drift of larval fish

The flow chart in Fig. 11.2 does not contain any de- 11.3.5 Adaptive traits cisions made by the organism. This situation may be relevant for the planktonic larval stages. Spatial As mentioned above, the strategy vector Si can be positioning of fish during this life stage is deter- used to specify the adaptive traits of an individual. mined by their buoyancy, the turbulence of the en- In some studies IBMs are simply tailored to match vironment and the advective transport of water. In the life history of target species (e.g. van Winkle et lakes, however, the impact of advective forces is al. 1993), in which case behaviour is not specified much smaller than in the sea and in rivers, where explicitly in the model, but the consequences of larval drift is an integrated part of the life cycle certain behavioural strategies for growth and of many fishes. Although assuming a spatially mortality may still be simulated. In other cases, homogenous compartment may be a reasonable finding the best behavioural trait is part of the simplification for some lakes, it is generally not so modelling exercise itself. We mainly discuss be- for ocean systems. haviour below, but life-history strategies can often Models for marine fishes have therefore in- be implemented in the same manner. In general creasingly been developed to incorporate advec- three approaches prevail for specifying behaviour tive fields and temperature data from ocean in IBMs: rule-based approaches, optimization, and circulation models (Bartsch et al. 1989; Werner et adaptation. al. 1996). These models are grid based with a reso- lution typically of 1–20km. Physical ocean circu- Rule-based approaches lation models provide current velocity vectors for each grid cell at each time step. These vectors are Even though rules may be chosen based on used to move individuals (or ‘particles’, repre- their evolutionary profitability, the essential part senting drifting biological organisms) about (e.g. of the rule-based approach is that the rules are 234 Chapter 11 provided by the modeller and that behaviour are well suited for situations where the entire life- results directly from the specified rule. One span of the animal is considered. However, it is example of a behavioural rule is random walk, more difficult to assess the profitability of behav- which simply is to move individuals about in iours at an instant. Solutions to this problem can a random fashion with equal probability of going be achieved using ‘rules of thumb’ that are gener- in each direction at each time step. Such a simple ated as predictions from evolutionary ecology, behavioural procedure resembles a natural situa- such as the ideal free distribution (Fretwell and tion in many cases of local searching in animals Lucas 1970), the marginal value theorem (Charnov (Berg 1993). Other rule-based concepts specified in 1976) and Gilliam’s rule (Werner and Gilliam models are the use of taxis and kinesis, which 1984). These concepts are applied to foraging and specify reactions to stimuli by orientating relative to habitat choice by Mittelbach (Chapter 11, Vol- to stimuli position or responding in proportion ume 1). Given the assumptions, these approaches to stimuli intensity respectively (Tyler and Rose will provide optimal solutions to the respective be- 1994). havioural problems.

Optimization Adaptation The second approach to modelling behaviour Adaptive models find ‘good’ behavioural strategies is generally referred to as the ‘optimization ap- by using gradual improvements in behaviour proach’ (Parker and Maynard Smith 1990). This through simulated evolution or learning. When approach underlies many of the discussions of simulating evolution, the behavioural strategies fish behaviour in Volume 1, as exemplified by are coded numerically on the strategy vector (equa- Mittelbach’s discussions of foraging theory in tion 11.4) and passed on from parents to offspring. Chapter 11, Volume 1. The optimization approach This technique is known as the genetic algorithm in behavioural ecology is ultimate, based directly (GA) (Holland 1975). In the GA a specific measure on the survival value of behavioural traits. Sto- of fitness such as R0 can be used to determine chastic dynamic programming (SDP) (Mangel and which strategy vectors and hence which individ- Clark 1988; Houston and McNamara 1999; Clark uals are the best, and these become parents for the and Mangel 2000), where the entire solution space new generation. Variability in the new strategy is sought and the best solution is chosen, is an ex- vectors are provided through recombinations ample of the optimization approach. SDP relates to among parents, and mutations (Holland 1975). As states rather than individuals and the optimal mentioned above, reinforcement learning can be strategy is calculated for each state (for example, used to update behaviours within the lifetime of weight or energy level) at each time step in a back- the individual (Ackley and Littman 1992). This ward iteration. This allows state variability and approach can also be used in combination with state-specific behaviour to emerge. Once the back- evolved strategies (Ackley and Littman 1992). ward iteration procedure is complete, an IBM can When the environment changes markedly on a be used to simulate individual trajectories from short time-scale, for example within generations, the start to the horizon of the model. SDP models it can be profitable to allow adaptation through may use Monte Carlo simulations for determining learning. In an artificial neural network (ANN) growth and reproduction, which makes this ap- model this would mean that the weights are proach individual-based (Clark and Mangel 2000). changed within the life of individuals, and not only Traditionally, population growth rates such as the between different generations. Artificial neural instantaneous rate of increase r or the lifetime networks (ANNs) apply neurobiological principles reproductive success R0 have been applied as fit- of synaptic brain activity to perform systematic ness measures in optimization models (Roff 1992; output by differential weighting of input variables Hutchings, Chapter 7, Volume 1). These measures (Rummelhart et al. 1986). Individual-based Models 235

Table 11.2 Recipe for incorporating behaviour and vorous fish are adapted over many consecutive spatial detail in an IBM of a planktivorous fish. The generations (Table 11.2). pseudo-code illustrates how the model is executed in a 1 The attribute and strategy vectors are initialized sequential manner, using a programming language such for each individual in the population. While the as FORTRAN. attribute vectors are set to ‘common’ values, the 1 Initiate attribute and strategy vectors for each individual strategy vectors are initiated randomly within cer- 2 Year loop tain intervals. The attribute vector is specified as: Create physical environment for the entire year: light, Ai(age, weight, energy level, position), and the cor- temperature, currents responding strategy vector is: Si(timing of spawn- Initiate food and predator distribution ing, spawning location, energy allocation rule, size 3 Day loop at maturity, movement). Movement behaviour is Update distribution and abundance of food, competitors and predators determined using an ANN and its weights are im- 4 Individual loop plemented in the strategy vector. Perform behavioural actions 2 At the start of the year the environment for the Determine growth and mortality coming year is established. As in the previous Update abundance of food items after feeding recipe, temperature and current fields are pro- Reproduce if criteria are fulfilled duced by an ocean circulation model. The predator abundance is assumed to increase linearly with in- creasing temperature. The food distribution is ini- tiated and is consistently updated below. An alternative approach for evolving behaviour 3 The day loop runs over each day of the year, and, in models is to use so-called emergent (endoge- most importantly, the food distribution is updated nous) fitness, which means that individuals with according to the production and import from ad- strategy vectors live and reproduce in an evolving vective transport provided by the physical model. population (Strand et al. in press). Thus rather than 4 In the individual loop most of the biological fea- maximizing a specific fitness measure, those indi- tures of the IBM are implemented. The first task is viduals who manage to pass on their own strategies to infer mortality. Since the model uses the super- by reproducing with other individuals in the popu- individual approach, a proportion of the number of lation will be the most fit. However, the degree to clones is removed according to the predation risk which this reflects a natural system depends upon at the present location. In the same fashion one the way the environment is specified in the par- may include removal of fish caused by fishing ticular model system. For applications of adaptive mortality. Food intake is determined from local models in ecology see below and Huse and Giske encounters with food, and growth is then calcu- (1998), Huse et al. (1999), and Strand et al. (in lated using the bioenergetic model of Hewett and press). Johnson (1992). For adults, surplus energy is divid- ed among growth and reproduction according to the individual allocation strategy. Juveniles, on Recipe for adaptive models with a strategy vector the other hand, are assumed to put all their energy IBMs containing both spatial detail and behaviour into growth. At an individually specified spawn- are generally very complex, since they often in- ing time and location, individuals may reproduce volve a range of mechanistic models for specifying given that certain criteria are fulfilled. The new temperature, drift, feeding and behaviour. The fol- individuals inherit behavioural and life history lowing recipe describes the conceptual flow of a strategies from their parents by recombination model containing both attribute and strategy vec- with a probability for mutations, to mimic biologi- tors and spatial detail. It is based on the model of cal reproduction. These are essential parts of the Huse (1998) and Huse and Giske (1998), where fish genetic algorithm. At the end of the individual behaviour and life history strategies for a plankti- loop, movement is determined using the ocean 236 Chapter 11 drift for the planktonic larvae (see recipe above) or involves problems that have high dimensionality the ANN for fish above a certain size. The ANN and/or include stochasticity or density depend- calculates movement from information about the encies, then the adaptive models will be the best local abundance of predators, growth, temperature approach. If one is interested in how large-scale and position. patterns and/or complex phenomena emerge from These points make up the basic structure of a spa- individual behaviour, it can be productive to apply tial life-history model of fish. The model is run for a simple behavioural rules. large number of years (300–500) and generates life- history and migration strategies that resemble ob- 11.3.6 Local interactions served migration patterns (Huse 1998). Although the example is for capelin migrations in the Barents In some cases, for example in fish schooling Sea, the model approach is general and can be ap- (Reynolds 1987; Vabø and Nøttestad 1997; Stöcker plied to virtually any fish stock. It has also been 1999), behaviour is dependent mainly upon applied successfully for simulating vertical migra- what conspecifics and predators in the vicinity are tion in mesopelagic fish (Huse et al. 1999; Strand doing. A simple modelling approach that takes et al. in press). local interactions into account is cellular auto- mata (for review see Phipps 1992). Under this ap- proach, the modeller defines strict rules that are What is the best modelling approach? similar for each individual cell in a lattice. The What type of model is ‘best’ for implementing rules then specify how the automata change state adaptive traits, or which one of these techniques according to the state of their surrounding cells. should one use? The answer is, as often – ‘it de- The emerging pattern of the lattice then results pends’. For a full discussion of individual-based from the local interactions among the automata. techniques see Tyler and Rose (1994) and Giske et al. (1998). Some general recommendations emerge from the discussion above. In terms of specifica- tion of adaptive traits in IBMs, the strategy vector 11.4 FORMULATING AND and the adaptation concept is appealing in many TESTING IBMS ways. The advantage of this concept is its gener- 11.4.1 Model formulation ality because it can encompass most ecological processes including density dependence, state de- Since IBMs typically are built from an extensive pendence and stochastic environments (Huse et al. set of submodels, there are many things that can go 1999; Strand et al. in press). Furthermore, when wrong during model formulation. It is therefore using ANNs it is possible to simulate behaviour important to develop a common framework for from stimuli, thereby allowing the use of conven- putting together IBMs. Railsback (2001) provides tional behavioural terminology and perspective. six points that should be considered when for- The downside of the adaptive approach is that it is mulating an IBM: (1) Emergence: what processes impossible to know whether the optimal solution should be imposed by empirical relations and what is found unless this is calculated by other means. should emerge from mechanistic representations? This is, however, ensured using optimality models (2) Adaptive traits: what kind of adaptive processes such as SDP, which is one of the great advantages of should be included in the model? This point has to that approach. Another advantage of SDP is the be related to the spatial scale and the major ques- ability of this technique to include individual state tions being addressed. (3) Fitness measure: what is and time constraints in the optimization criterion. the appropriate fitness measure for the adaptive A conclusion can therefore be that if the biological traits of the model? (4) State-based dynamics: how question involves state dependence or making should decision processes depend on individual sure that the optimal strategy is found, then SDP state? (5) Prediction: what are realistic assump- should be used. On the other hand, if the study tions about how animals predict the consequences Individual-based Models 237 of decisions? (6) Computer implementation: what ondary model predictions. With regard to a fish user interface is necessary for implementing, model this can be, for example, testing the perfor- validating and testing IBMs? In addition to these mance of a bioenergetic model used, even though points it is important to provide results that can the primary aim of the model is to provide popula- be tested against observations. Although testing tion dynamics of a target species. The primary and against real data is not always necessary, since secondary model predictions can then be tested models may provide valuable insight through sen- independently of each other. A consequence of sitivity analyses, it is most often an advantage in model evaluation is either to accept the model’s model development. The points shown above are performance or, alternatively, to try to revise parts important to keep in mind when constructing an of the model that produce erroneous predictions. IBM, and we shall return to some of these as we go One should, however, be careful not to make the along. Other things to keep in mind when formu- model fit observations by changing parameter val- lating a model are that the choice of complexity ues or submodels uncritically. This is especially and model structure should be based on the level of important for IBMs, which typically simulate understanding of the environmental and biologi- processes in a mechanistic manner using a large cal processes operating. Lastly a model is at best a number of parameter values. highly simplified but biased representation of na- When it comes to software implementation, ture, and uncertainty can be reduced by attacking any kind of programming language can be used, but the problem with several models that differ in as- object-orientated languages such as C++ are espe- sumptions and structure (see also Sparre and Hart, cially well suited for constructing IBMs (Maley Chapter 13, this volume). and Caswell 1992). In addition to providing a nice structured programming with individuals as the 11.4.2 Evaluation of IBMs basic units, these languages generally also provide good visualization opportunities. For example, the Model evaluation is an important part of model de- object-orientated Swarm package (Langton et al. velopment, and for full descriptions of this process 1999) is tailored for individual-based simulations see Jørgensen (1988) and Bart (1995). The evalua- and provides a number of features for visualization tion process can be divided into verification and of results and bookkeeping of individuals and validation (Jørgensen 1988). Verification is the processes. Using Swarm, several aspects of the process of checking that the internal logic of the model may be monitored during development and computer model is correct and that the model ac- evaluation. Swarm, which is a shareware product, tually does what it is intended to do. This process is is currently available for the Objective C and Java performed continuously as a model is developed. languages. Another software package especially Validation, on the other hand, aims at determining developed for individual-based simulations is the ability of the model in describing observed ECOSIM (Lorek and Sonnenschein 1998). This is phenomena. This process can be divided further not the same as the Ecosim discussed by Pauly and into checking the validity of: parameter values, Christensen (Chapter 10, this volume). and secondary and primary model predictions (Bart 1995). The validity of parameter values is tra- ditionally tested through a sensitivity analysis. 11.5 REVIEW OF Sensitivity analyses involve varying parameter INDIVIDUAL-BASED MODELS values and studying the effect on model output. IN FISHERIES BIOLOGY For IBMs, which usually have a great number of parameters, it is simply not feasible to test the sen- The ontogenetic development in fishes from eggs sitivity of all these. Rather some ‘key’ parameters to maturity typically involves manyfold increases should be chosen for testing. Since IBMs tend to be in body size associated with discrete changes composed of submodels, many features of an IBM in morphology and increased behavioural reper- can be tested. This is what is referred to as sec- toires. The way individuals vary therefore changes 238 Chapter 11

Table 11.3 Features characterizing different aspects of the early life history of fish.

What Variables References characterizes . . .

the survivors? egg quality Kjesbu et al. (1991) birth date Schultz (1993) birth position Berntsen et al. (1994); Slotte and Fiksen (2000) egg size Knutsen and Tilseth (1985) development Blaxter (1986) prey encounter Fiksen and Folkvord (1999) bioenergetics Crowder et al. (1992); Fiksen and Folkvord (1999) the environment? food concentration Cushing (1990); (1996) small-scale turbulence Sundby and Fossum (1990); MacKenzie et al. (1994) water Bartsch et al. (1989); Berntsen et al. (1994); transportation Hermann et al. (1996) ocean climate Cushing (1996); Anderson and Piatt (1999) light Miner and Stein (1993); Fiksen et al. (1998) turbidity Chesney (1989); Fortier et al. (1996) temperature Houde (1989); (1997) predators McGurk (1986); Bailey and Houde (1989); Cowan et al. (1996)

through the life cycle. This suggests that model- considered recruitment variation as a major re- ling should focus on different aspects of life at search area in fisheries science (Myers, Chapter 6, different times of the life cycle, and that each onto- Volume 1). Recruitment variation may be caused genetic stage might be specified differently. For ex- by factors related to the state of the organism, its ample, in the period of change from intrinsic to environment or its parents (Table 11.3; see Myers, extrinsic energy uptake, each prey caught by a lar- Chapter 6, Volume 1). IBMs allow a mechanistic val fish is very important, which makes food gath- representation of ecological processes, and conse- ering a vital process to simulate in models. During quently such models have been applied exten- overwintering or spawning, on the other hand, sively in studying growth and survival of feeding will be a virtually unimportant activity. In young-of-the-year fish (see review by van Winkle the following we will review studies using IBMs in et al. 1993). Beyer and Laurence (1980) recognized fish biology by using the structure of Fig. 11.1. the importance of initial chance events in feeding Hence we initially discuss relatively simple mod- of larval winter flounder. Since prey ingestion can els, and then add features to the IBMs as we go be described as a Poisson process, some individuals along. The first paragraph therefore focuses on will initially have success and capture prey while growth and survival in early life-history studies, others have poor fate. Given that the ability to ob- whereas the later paragraphs discuss studies of tain food is size dependent (Juanes et al., Chapter greater relevance to older life stages. 12, Volume 1), there will be a positive feedback so that initially successful individuals will grow 11.5.1 Models of growth and survival faster than the others and thus have a higher sur- vival probability. Such scenarios are well repre- For most, if not all, fish species, mortality at the sented in IBMs (Beyer and Laurence 1980). egg and early larval stages is enormous compared The important survival factors for eggs and to later in life. Therefore, as early as 1914, Hjort larvae are prey and predator encounters, bioener- Individual-based Models 239

(a) (b)

Fig. 11.4 The effect of initial variation in size distribution on the outcome of cannibalistic (c) (d) interactions. Low initial variance Abundance (a) produces a cohort with homogeneous individuals at a later stage (b), while with a high initial variance (c) the biggest individuals get cannibalistic (d). (From Huston et al. 1988; reproduced by permission of BioScience.) Size Size

getics and organismal development. With so many initial size were able to become cannibalistic and factors interacting, modelling is essential in eat the smaller fish, which the model run predicted understanding how this affects individuals and to lead to survival of few but large individuals (Fig. populations (Houde 1997). And since egg and larval 11.4d). It was demonstrated that the individual mortality are high, models of survival should be variability is clearly important for cohort develop- able to focus on the lucky or clever few. The impor- ment even though the initial average weight was tant decision variables are mainly governed by the same in the two cases. The model presented by parent behaviour through choice of spawning site Fiksen and Folkvord (1999) provides a state-of-the- and time, anatomical properties such an egg size art mechanistic description of the feeding process and energy density, and by developmental ‘pro- taking into account environmental features such gramme’ displayed through the sequence of organ as small-scale turbulence, light, turbidity, tem- development (Jobling, Chapter 5, Volume 1). A perature, prey density and size structure. Along- range of ecological conditions will bring stochas- side the model development there has been ticity to all these variables (Table 11.3). The reli- experimental work used for generating parameter ability of predictions from a complex recruitment values and testing of model predictions. The great model will depend on how well these processes are detail of the mechanistic description enables this represented mechanistically (Fiksen and Folkvord model to make more realistic predictions than 1999). In a classic study, DeAngelis et al. (1979) models with simpler environmental description. modelled the growth of young-of-the-year large- mouth bass. They found that the initial variance in 11.5.2 Models with adaptive traits the cohort was important for its development and for whether cannibalism was possible or not. In the The simplest behavioural models are those with- case of low initial variance, shown by an even size out spatial detail. There have been some studies distribution (Fig. 11.4a), the fish remained homo- addressing how spawning strategies are affected geneous with none growing big enough to eat the by seasonal variation in temperature, growth and others (Fig. 11.4b). With large initial variance (Fig. predation risk. Trebitz (1991) used an IBM to find 11.4c), however, the individuals that had largest the best timing and temperature for spawning with 240 Chapter 11 and without density dependence. Initially the cell to stay and/or reproduce. LePage and Cury spawning temperature was given randomly, but (1997) used SeaLab to simulate how reproductive subsequent spawning strategies (temperature) strategies depended on the degree of variation in were given in proportion to the survival of the variable environments. An obstinate strategy, strategies (individuals) until age 1. The biomass where spawning is performed under the same con- of each temperature strategy at age 1 was there- ditions as the individual is born, and an opportunis- fore used as a measure of strategy performance. tic strategy where individuals spawn under novel Density-independent and density-dependent environmental conditions, were investigated. For model runs predicted different optimal spawning extreme environmental variation, only popula- temperatures, with a broad peak at intermediate tions with both the opportunistic and obstinate temperatures for the density-independent situa- strategies survived. The authors discuss the results tion and an almost temperature-independent with regard to straying in natural populations. A profitability for the density-dependent case. This similar artificial environment was applied by model is in many ways similar to the genetic algo- Anneville et al. (1998), who explored the effect rithm discussed above. of density-dependent recruitment relationships using rule-based movement in an hexagonal lat- 11.5.3 Simple spatial systems tice. Their main conclusion was that local density dependence was often not detectable at large spa- Natural systems are very complex. It may therefore tial scales, which stresses the importance of being be profitable to conduct simulation experi- explicit about scale when analysing ecological ments, which is a computer analogy to laboratory processes (Levin 1992). or field experiments, in which focus can be put on a specific research topic without necessarily 11.5.4 Models of fish distribution defining a particular natural system. A simulation experiment is especially fruitful early in theoreti- Spatially explicit IBMs incorporate spatial hetero- cal developments or if the problem is general rather geneity, individual variability and individual than linked to a specific population or environ- movement (Tyler and Rose 1994). Although such ment. Tyler and Rose (1997) studied cohort conse- models are complicated and necessarily consist quences of different habitat choice rules in a simple of many modules, they provide the potential for spatial system using life-history-based criteria highly realistic simulations of fish populations. such as Gilliam’s rule. Their results suggest that Many fishes undertake extensive horizontal mi- habitat choice rules have strong effects on cohort grations between feeding, spawning and over- survivorship and that no single departure rule can wintering areas (Metcalfe et al., Chapter 8, Volume be an evolutionarily stable strategy. They therefore 1). Such migrations involve complex interactions concluded that the use of static life-history rules between individuals and the environment that are is not an appropriate way to model behaviour in generally not well understood. Tagging experi- a dynamic environment, as confirmed elsewhere ments have provided first-hand information about (Railsback et al. 1999). Tyler and Rose (1997) also the distribution of migratory fish stocks, and lately found that density-dependent effects on juvenile telemetric tags have enabled real-time observa- survival can be much greater in spatially explicit tions of fish movements in the sea (Metcalfe and models with fitness-based habitat choice than in Arnold 1997; Metcalfe et al., Chapter 8, Volume 1). spatially homogeneous models. SeaLab (LePage However, it is difficult to understand the mecha- and Cury 1997) is a spatial simulator that can be nisms controlling the movement of fish just from used to test hypotheses about fish reproduction and observations, and model simulations are therefore space use. This model divides space into hexagonal an important part of exploring the proximate and structures, and individuals may choose in which ultimate processes involved in fish migrations. Individual-based Models 241

Although there is likely to be some common the Barents Sea, and assumed distributions of features among fish stocks in how they move hori- predators and prey. By using a time step of one zontally, local adaptation will be important, and it month, the optimal habitat for each state (body may not be trivial to transfer knowledge about the weight) was calculated using the SDP equations migration of one stock to other stocks. In the case with lifetime reproductive success as a fitness of vertical migration, however, there may be more criterion. The model results compared favourably common features due to the pronounced vertical with the observed distribution of the Barents Sea gradients and diurnal changes in light that strongly capelin. In contrast to Walter et al. (1997), Fiksen influence visual feeding and predation risk. et al. (1995) assumed that the individual capelin Larval drift was among the first topics where actively made habitat choices to maximize their spatially explicit IBMs were applied. Werner et al. Darwinian fitness. Huse and Giske (1998) and (1996) considered trophodynamics and ocean cir- Huse (1998) used a similar environmental descrip- culation in a study of cod larvae on Georges Bank tion as Fiksen et al. (1995) to simulate movement by providing drift and growth trajectories of indi- of the Barents Sea capelin using an adaptive model viduals. They concluded that the region of highest with a strategy vector (equation 11.4) containing retention coincided with the region of highest the weights of an artificial neural network and growth, illustrating the complementary interac- several life-history traits. Similarly to Fiksen et al. tion between trophodynamics and circulation (1995), they assumed that the capelin is adapted processes. In a similar model of walleye pollock to the environment of the Barents Sea. However, (Theragra chalcogramma), Hinckley et al. (1996) rather than assuming optimal habitat choices, an found that the inclusion of mechanisms that adaptive process over many hundreds of genera- determine the depth positioning of the drifting tions was simulated using emergent fitness (see larvae are important for determining the direction above). The model was applied to study evolution of horizontal advection. of spawning areas, and it predicted, in accordance Through a combination of simple decision with field observations, capelin spawning to occur rules and a library of surface currents, Walter et al. along the coast of Northern Norway. ANN-based (1997) modelled sockeye salmon (Onchorhynchus models may rely on proximate sensor information, nerka) movement in the Northeast Pacific Ocean. and at the same time the weights of the ANN are Movement was simulated based on random walk, adapted using ultimate forces. In this way ANN with directed swimming at certain times of the models provide a link between proximate and year. Their simple rules for compass bearing pre- ultimate factors in behavioural ecology. ANNs dict migration patterns that challenge prevailing have also been applied to study tropical tuna complex models of sockeye migration. The most (Katsuwonus pelamis and Thunnus albacares) striking result was that inter-annual variation in migrations between the Mozambique Channel the surface current led to great changes in the and the Seychelles using temperature data gath- distribution of the salmon given that the same ered by remote sensing (Dagorn et al. 1997). In this movement rules apply each year. The rule-based case the tuna was assumed to search for areas of approach used by Walter et al. (1997) does not low temperature, which are often associated with take into account that individual differences in food-rich frontal zones. The ANN was used to de- state can cause differences in behaviour. State- termine which movement action the artificial dependent habitat choice was considered by tuna should make based on information about the Fiksen et al. (1995), who used a stochastic dynamic temperature map within the daily search radius. programming (SDP) to simulate the horizontal The fitness criterion used in the GA for training distribution of the Barents Sea capelin (Mallotus the ANN was to minimize the distance to the ob- villosus). The model was based on prior simulation served arrival point near the Seychelles. As a result results, which specify the physical environment of of the adaptive process, the simulated tuna eventu- 242 Chapter 11 ally managed to get from the starting point in the to coordinate their behaviour during predator Mozambique Channel to the Seychelles. In this attacks. model the tuna is assumed only to consider forag- Stöcker (1999) used cellular automata to study ing potential when making habitat choices and not energetic aspects of schooling. The results were to consider predation, as in the two studies on discussed with regard to tuna, for which saving capelin mentioned above. Nevertheless, because energy is believed to be an important motiva- adult tuna are large they tend to have a low risk of tion for school formation. The model predicts predation compared with smaller fish, and forag- school break-ups due to oxygen depletion, and ing is therefore the key aspect of tuna migrations can thus be used to estimate maximum school outside the spawning season. sizes for tuna. Both Vabø and Nøttestad (1997) and Stöcker (1999) assumed identical rules for each 11.5.5 Local interactions: schooling individual. Romey (1996), on the other hand, in another schooling model showed that individuals Schooling consists of behaviours where coordina- with biased behaviour compared with the re- tion between individuals is important. As differ- maining group had a large influence on school ent from more loosely organized shoals, schools behaviour. The degree to which individual differ- are defined as compact, coordinated and polarized ences in behaviour impacts school dynamics collections of fish (Pitcher and Parrish 1993; may be a function of school size, with a greater Krause et al., Chapter 13, Volume 1). Schooling impact of individual variability in small schools. is a characteristic phenomenon of many fishes, The cellular automata technique is well suited explained mainly as a behavioural feature for for simulating the dynamics of schools with predatory defence, but it also acts as a collective similar individual behaviour, but the IBM models mechanism for coordinating migration and repro- developed by Reynolds (1987) and Vabø and duction (Pitcher and Parrish 1993). Even though Nøttestad (1997) are more applicable than cellular the ultimate causes of schooling are well known, automata if individual differences in behaviour are the sensory and behavioural mechanisms involved believed to be important. For a general discus- in coordination and control of schooling remain sion of predictions made by cellular automata largely unresolved. Schooling relies on the co- and a conventional IBM respectively see Lett et al. ordinated movement of individuals, and school (1999). models are usually based on the assumption that individuals follow similar behavioural rules, 11.5.6 Responsive environments as opposed to most IBMs which allow individual and natural stock sizes variability. Reynolds (1987) simulated general flocking behaviour, which includes schooling, in a Most of the models above describe individuals in seminal paper on collective behaviour. He showed non-responsive environments. This means that that by using three simple individual rules, includ- individuals are not assumed to impact on their ing matching of the movement vector of neigh- environment, which is unrealistic when density bours, staying close to others, and avoiding contact dependence is important. Huse and Giske (1998) with other individuals and obstacles, flocking be- applied the super-individual approach to model haviour that resembles observations of birds long-distance migration of the Barents Sea capelin. could be recreated. Vabø and Nøttestad (1997) used The number of super-individuals was limited a similar rule-based IBM to simulate the behaviour to a maximum of 15000, and at birth each super- of herring schools (Clupea harengus) when at- individual represented one million identical tacked by a killer whale (Orcinus orca). The siblings. Depletion of zooplankton from feeding model was able to recreate most of the avoidance and consequent relocation is important for under- manoeuvres seen in herring, and the results standing capelin distribution dynamics (Hassel thus suggest that herring may use simple rules et al. 1991), and by using this approach it was possi- Individual-based Models 243 ble to simulate realistic stock sizes of capelin be profitable to use models that simulate both the while incorporating trophic feedback. The model dynamics of the parents as well as those of the re- predicts that fish should be more dispersed in cruiting larvae. Such models potentially may yield model runs with trophic feedback compared to early forecasts of stock recruitment, which is part those where food was not removed as a result of of the management process. feeding by the fish, and hence illustrates the Since different age and size groups of fish often importance of including trophic feedback even at exhibit different behaviours, spatial dynamics and large spatial scales. reproductive capabilities, they impact the stock dynamics in different ways. How fishing effort is 11.5.7 Applications of IBMs in distributed over the different size groups and spa- fisheries assessment tial areas may thus affect the population response to harvest (Sparre and Hart, Chapter 13, this While structured models, such as the virtual popu- volume). IBMs are well suited to simulate the ef- lation analysis and related approaches, are com- fects that management decisions can have both on mon in fisheries assessment, IBMs have not been population dynamics in the short run and on evo- used for this purpose to any degree. Most applica- lutionary dynamics in the longer run (Martinez- tions of IBMs have focused on ecological questions Garmendia 1998). of general interest with the aim of exploring IBMs can also be used in fisheries assessment to how individual variability in state and strategies facilitate the abundance estimation process, often influence the dynamics at the population level. performed by acoustics, sampling, or a combina- Improved understanding of the causes of temporal tion of these. One potential application of IBMs is and spatial variation in mortality, growth and to predict distribution patterns of fish populations recruitment, but also of migration and distribu- in the ocean given a certain environmental regime, tion patterns, has been achieved. Along with the which can allow improved temporal and spatial ongoing increase in computing power, new possi- survey coverage. IBMs can also be applied to take bilities for individual-based approaches to man- into account the behavioural response of fish to agement problems of fish populations can be survey vessels and sampling equipment. Diving possible. In order for IBMs to be used for manage- behaviour and changes in tilt of fish can strongly ment purposes it will be essential that the IBMs impact on its echo-reflecting properties from the are robust and reliable with regard to structural swim bladder of the fish (Huse and Ona 1996; Brix, assumptions, parameter values, primary and Chapter 4, Volume 1), and hence impact on the secondary model predictions (Bart 1995). These acoustic estimate obtained. If such behaviours factors reflect the general recommendations for can be predicted and corrected for, the quality of formulating and evaluating IBMs as discussed acoustic estimates may be improved. This can be above. IBMs have been applied to some degree in approached using ANNs, where observations of combination with bioenergetics models for man- fish behaviour can be used to train the network. By agement purposes (Hansen et al. 1993). IBMs can presenting many sets of observed behaviours and also be fruitful for management purposes in at relation to the vessel such as angle, depth and dis- least four other areas: (1) in predicting stock re- tance, the network can be trained using these cruitment, (2) predicting the response of different observations to generalize responses of fish to exploitation patterns on the stock, (3) by facili- vessel presence. The response to trawls and other tating the abundance estimation process and (4) sampling equipment can be simulated in a similar through estimating stock prognoses in spatially fashion. explicit models. IBMs can also be used to predict population dy- Detailed IBMs (e.g. Fiksen and Folkvord 1999) namics of stocks through spatially explicit models can be used to elucidate which processes control covering the entire target stock. Such models have recruitment to a target stock. For example, it may not made an impact on stock assessment. How- 244 Chapter 11 ever, it can be a potential future application area of REFERENCES IBMs, and the modelling approaches of Fiksen et al. (1995) and Huse and Giske (1998) are attempts to Ackley, D. and Littman, M. (1992) Interactions between move in this direction. learning and evolution. In: C. Langton, C. Taylor, J. Farmer and S. Rasmussen (eds) Artificial Life III. Reading, MA: Addison-Wesley, pp. 487–509. 11.6 CONCLUSIONS Anderson, P.J. and Piatt, J.F. (1999) Community reorgani- zation in the Gulf of Alaska following ocean climate Even though IBMs are yet to have made an impact regime shift. Marine Ecology Progress Series 189, 117–23. on fisheries management, the modelling tech- Anneville, O., Cury, P., LePage, C. and Treuil, J.P. (1998) nique has become especially popular among fish Modelling fish spatial dynamics and local density- biologists (Grimm 1999). The reason for this popu- dependence relationships: detection of patterns at a larity is partly historical, as some of the first appli- global scale. Aquatic Living Resources 11, 305–14. cations of IBMs (DeAngelis et al. 1979; Beyer and Asplin, L., Salvanes, A.G.V. and Kristoffersen, J.B. (1999) Laurence 1980), were in fish biology. But as impor- Nonlocal wind-driven fjord-coast advention and its potential effect on plankton and fish recruitment. tant is the great interest in explaining the observed Fisheries Oceanography 8, 255–63. recruitment variability seen in most fish species Bailey, K.M. and Houde, E.D. (1989) Predation on the eggs (Grimm 1999). IBMs are particularly useful for and larvae of marine fishes and the recruitment prob- studying recruitment variability, but in addition lem. Advances in Marine Biology 25, 1–83. the approach allows features such as behaviour Bart, J. (1995) Acceptance criteria for using individual- and life histories to be studied as discussed based models to make management decisions. Ecologi- above. IBM predictions are generally easy to cal Applications 5, 411–20. Bartsch, J. and Knust, R. (1994) Simulating the dispersion compare with individual observations, which is of vertically migrating sprat larvae (Sprattus sprattus another advantage of the approach. The use of L.) in the German Bight with a circulation and trans- super-individuals (Scheffer et al. 1995) allows port model system. Fisheries Oceanography 3, 92–105. simulation of realistic fish stock abundances Bartsch, J., Brander, K., Heath, M., Munk, P., Richardson, while maintaining an individual-based modelling K. and Svendsen, E. (1989) Modelling the advection of structure. IBMs are often very complex and com- herring larvae in the North Sea. Nature 340, 632–6. posed by many submodels. Model evaluation is Berg, H.C. (1993) Random Walks in Biology. Princeton, NJ: Princeton University Press. therefore very important, and preferably both pri- Berntsen, J., Skagen, D.W. and Svendsen, E. (1994) Model- mary and secondary model predictions should be ing the drift of particles in the North Sea with reference validated (Bart 1995). Whether or not IBMs really to sandeel larvae. Fisheries Oceanography 3, 81–91. give different answers to ecological problems than Beyer, J.E. and Laurence, G.C. (1980) A stochastic model do the traditional state-variable models remains to of larval fish growth. Ecological Modelling 8, 109–32. be seen, but nevertheless, IBMs provide modellers Blaxter, J.H.S. (1986) Development of sense organs and behaviour of teleost larvae with special reference to with a highly flexible tool for studying individuals feeding and predator avoidance. Transactions of the and populations. American Fisheries Society 115, 98–114. Carlotti, F., Giske, J. and Werner, F. (2000) Modeling zooplankton dynamics. In: R. Harris, P. Wiebe, J. Lenz, ACKNOWLEDGEMENTS H.R. Skjoldal and M. Huntley (eds) ICES Zooplankton Methodology Manual. London: Academic Press, pp. We thank Steve Railsback and David Kirby 571–667. Caswell, H. and John, A.M. (1992) From the individual to for fruitful comments on an earlier draft of this population in demographic models. In: D.L. DeAngelis chapter. Anne Gro Vea Salvanes and Geir Huse and L.J. Gross (eds) Individual-based Models and were supported by the Research Council of Approaches in Ecology. New York: Chapman & Hall, Norway. pp. 36–66. Individual-based Models 245

Caswell, H. (1996) Matrix methods for population analy- DeAngelis, D.L., Rose, K.A., Crowder, L.B., Marshall, sis. In: S. Tuljapurkar and H. Caswell (eds) Structured E.A. and Lika, D. (1993) Fish cohort dynamics – appli- Population Models in Marine, Terrestrial, and Fresh- cation of complementary modelling approaches. The water Systems. New York: Chapman & Hall. American Naturalist 142, 604–22. Chambers, R.C. (1993) Phenotypic variability in fish Fiksen, Ø. and Folkvord, A. (1999) Modelling growth and populations and its representation in individual-based ingestion processes in herring Clupea harengus larvae. models. Transactions of the American Fisheries Marine Ecology Progress Series 184, 273–89. Society 122, 404–14. Fiksen, Ø., Giske, J. and Slagstad, D. (1995) A spatially Charnov, E.L. (1976) Optimal foraging: the marginal explicit fitness-based model of capelin migrations the value theorem. Theoretical Population Biology 9, Barents Sea. Fisheries Oceanography 4, 193–208. 129–36. Fiksen, Ø., Utne, A.C.W., Aksnes, D.L., Eiane, K., Chesney, E.J. (1989) Estimating the food requirements of Helvik, J.V. and Sundby, S. (1998) Modeling the influ- striped bass larvae Morone saxatilis: effects of light, ence of light, turbulence and development on foraging turbidity and turbulence. Marine Ecology Progress in larval cod and herring. Fisheries Oceanography 7, Series 53, 193–200. 355–63. Clark, C.W. and Mangel, M. (2000) Dynamic State Vari- Fortier, L., Gilbert, M., Ponton, D., Ingram, R.G., able Models in Ecology: Methods and Applications. Robineau, B. and Legendre, L. (1996) Impact of fresh Oxford: Oxford University Press. water on a subarctic coastal ecosystem under seasonal Cowan, J.H., Houde, E.D. and Rose, K.A. (1996) Size- ice (southeastern Hudson Bay, Canada): III. Feeding dependent vulnerability of marine fish larvae to preda- success of marine fish larvae. Journal of Marine Sys- tion: An individual-based numerical experiment. ICES tems 7, 251–65. Journal of Marine Science 53, 23–37. Fretwell, S.D. and Lucas Jr, H.J. (1970) On territorial Crowder, L.B., Rice, J.A., Miller, T.J. and Marschall, E.A. behavior and other factors influencing habitat dis- (1992) Empirical and theoretical approaches to size- tributions in birds: 1. Theoretical development. Acta based interactions and recruitment variability in Biotheoretica 19, 16–36. fishes. In: D.L. DeAngelis and L.J. Gross (eds) Giske, J., Huse, G. and Fiksen, Ø. (1998) Modelling Individual-based Models and Approaches in Ecology. spatial dynamics of fish. Reviews in Fish Biology and New York: Chapman & Hall, pp. 237–55. Fisheries 8, 51–91. Cushing, D.H. (1990) Plankton production and year- Grimm, V. (1999) Ten years of individual-based model- class strength in fish populations: an update of the ling in ecology: what have we learned and what could match/mismatch hypothesis. Advances in Marine we learn in the future? Ecological Modelling 115, Biology 26, 249–93. 129–48. Cushing, D.H. (1996) Towards a science of recruit- Hansen, M.J., Boisclair, D., Brandt, S.B., Hewett, ment in fish populations. In: O. Kinne (ed.) Excellence S.W., Kitchell, J.F., Lucas, M.C. and Ney, J.J. (1993) in Ecology, vol. 7. Oldendorf/Luhe: Ecology Institute. Applications of bioenergetics models to fish ecology Dagorn, L., Petit, M. and Stretta, J.M. (1997) Simulation and management – where do we go from here? Trans- of large-scale tropical tuna movements in relation actions of the American Fisheries Society 122, with daily remote sensing data: the artificial life ap- 1019–30. proach. Biosystems 44, 167–80. Hassel, A., Skjoldal, H.R., Gjøsæter, H., Loeng, H. and Darwin, C. (1859) The Origin of Species. [1968], New Omli, L. (1991) Impact of grazing from capelin York: Penguin Books. (Mallotus villosus) on zooplankton: a case study in the DeAngelis, D.L. and Gross, L.J. (eds) (1992) Individual- northern Barents Sea in August 1985. Polar Research based Models and Approaches in Ecology. New York: 10, 443–60. Chapman & Hall. Hermann, A.J., Hinckley, S., Megrey, B.A. and Stabeno, DeAngelis, D.L. and Rose, K.A. (1992) Which individual- P.J. (1996) Interannual variability of the early life his- based approach is most appropriate for a given prob- tory of walleye pollock near Shelikof Strait as inferred lem? In: D.L. DeAngelis and L.J. Gross (eds) from a spatially explicit, individual-basd model. Fish- Individual-based Models and Approaches in Ecology. eries Oceanography 5, 39–57. New York: Chapman & Hall, pp. 67–87. Hewett, S.W. and Johnson, B.J. (1992) An Upgrade of a DeAngelis, D.L., Cox, D.C. and Coutant, C.C. (1979) Generalized Bioenergetics Model of Fish Growth for Cannibalism and size dispersal in young-of-the-year Microcomputers. University of Wisconsin, Wisconsin largemouth bass: experiments and model. Ecological Sea Grant College Program, Sea Grant Technical Modelling 8, 133–48. Report, WIS-SG-92-250, Madison. 246 Chapter 11

Hinckley, S., Hermann, A.J. and Megrey, B.A. (1996) De- Langton, C.G., Minar, N., Burkhart, R., Askenazi, M. and velopment of a spatially explicit, individual-based Ropella, G. (1999) The Swarm simulation system. model of marine fish early life history. Marine Ecology Web-based documentation at http://www.swarm.org. Progress Series 139, 47–68. Le Page, C. and Cury, P. (1997) Population viability and Hjort, J. (1914) Fluctuations in the great fisheries of spatial fish reproductive strategies in constant and northern Europe reviewed in the light of biological changing environments: an individual-based model- research. Rapports et Procès-Verbaux des Réunions du ling approach. Canadian Journal of Fisheries and Conseil International pour l’Exploration de la Mer 20, Aquatic Sciences 54, 2235–46. 1–28. Lett, C., Silber, C. and Barret, N. (1999) Comparison of Holland, J.H. (1975) Adaptation in Natural and Artificial cellular automata network and an individual-based Systems. Ann Arbor: University of Michigan Press. model for the simulation of forest dynamics. Ecologi- Houde, E.D. (1989) Comparative growth, mortality and cal Modelling 121, 277–93. energetics of marine fish larvae: temperature and im- Levin, S.A. (1992) The role of space in ecology. Ecology plied latitudinal effects. Fishery Bulletin 87, 471–95. 73, 1943–67. Houde, E.D. (1997) Patterns and consequences of selec- Levins, R. (1969) Some demographic and genetic conse- tive processes in teleost early life histories. In: R.C. quences of environmental heterogeneity for biological Chambers and E.A. Trippel (eds) Early Life History and control. Bulletin of the Entomological Society of Recruitment in Fish Populations, Fish and Fisheries America 15, 237–40. Series 21. London: Chapman & Hall. L¢ omnicki, A. (1988) Population Ecology of Individuals. Houston, A.I. and McNamara, J.M. (1999) Models of Princeton, NJ: Princeton University Press. Adaptive Behaviour. Cambridge: Cambridge Univer- Lorek, H. and Sonnenschein, M. (1998) Object oriented sity Press. support for modelling and simulation of individual- Huse, G. (1998) Life history strategies and spatial dynam- oriented ecological models. Ecological Modelling 108, ics of the Barents Sea capelin (Mallotus villosus). PhD, 77–96. thesis, University of Bergen. Lynch, D.R., Gentleman, W.C., McGillicuddy, D.J. and Huse, G., Strand, E. and Giske, J. (1999) Implementing Davis, C.S. (1998) Biological/physical simulations of behaviour in individual-based models using neural Calanus finmarchicus population dynamics in the networks and genetic algorithms. Evolutionary Ecol- Gulf of Maine. Marine Ecology Progess Series 169, ogy 13, 469–83. 189–210. Huse, G. and Giske, J. (1998) Ecology in Mare Pentium: MacKenzie, B.R., Iller, T.J., Cyr, S. and Leggett, W.C. an individual based spatio-temporal model for fish (1994) Evidence for a dome-shaped relationship be- with adapted behaviour. Fisheries Research 37, tween turbulence and larval fish ingestion rates. Lim- 163–78. nology and Oceanography 39, 1790–9. Huse, I. and Ona, E. (1996) Tilt angle distribution and Magurran, A. (1986) Individual differences in fish behav- swimming speed of overwintering Norwegian spring iour. In: T.J. Pitcher (ed.) The Behaviour of Teleost spawning herring. ICES Journal of Marine Science 53, Fishes. Baltimore, MD: The Johns Hopkins University 863–73. Press, pp. 338–65. Huston, M., DeAngelis, D.L. and Post, W. (1988) New Maley, C.C. and Caswell, H. (1992) Implementing i-state computer models unify ecological theory. BioScience configuration models for population dynamics – an 38, 682–91. individual based approach. Ecological Modelling 68, Jørgensen, S.E. (1988) Fundamentals of Ecological 75–89. Modelling. Amsterdam: Elsevier. Mangel, M. and Clark, C.W. (1988) Dynamic Modeling in Judson, O. (1994) The rise of individual-based models in Behavioral Ecology. Princeton, NJ: Princeton Univer- ecology. Trends in Ecology and Evolution 9, 9–14. sity Press. Kjesbu, O.S., Klungsøyr, J., Kryvi, H., Witthames, Martínez-Garmendia, I. (1998) Simulation analysis of P.R. and Walker, M.G. (1991) Fecundity, atresia, and evolutionary response of fish populations to size selec- egg size of captive Atlantic cod (Gadus morhua) in tive harvesting with the use of an individual based relation to proximate body-composition. Canadian model. Ecological Modelling 111, 37–60. Journal of Fisheries and Aquatic Sciences 48, 2333– McGurk, M.D. (1986) Natural mortality of marine 43. pelagic fish eggs and larvae: role of spatial patchiness. Knutsen, G.M. and Tilseth, S. (1985) Growth, develop- Marine Ecology Progress Series 34, 227–42. ment and feeding success of Atlantic cod (Gadus McQuinn, I.H. (1997) Metapopulations and the Atlantic morhua) larvae related to egg size. Transactions of the herring. Reviews in Fish Biology and Fisheries 7, American Fisheries Society 114, 507–11. 297–329. Individual-based Models 247

Metcalfe, J.D. and Arnold, G.P. (1997) Tracking fish with Scheffer, M., Baveco, J.M., DeAngelis, D.L., Rose, K.A. electronic tags. Nature 387, 665–6. and van Nes, E.H. (1995) Super individuals: a simple Metz, J.A.J. and Diekman, O. (1986) The Dynamics of solution for modelling large populations on an indivi- Physiologically Structured Populations. Berlin: dual basis. Ecological Modelling 80, 161–70. Springer-Verlag. Schultz, E.T. (1993) The effect of birth data on fitness of Miner, J.G. and Stein, R.A. (1993) Interactive influence female dwarf perch, Micrometrus minimus. Evolution of turbidity and light on larval bluegill (Lepomis 47, 520–39. macrochirus) foraging. Canadian Journal of Fisheries Slotte, A. and Fiksen, Ø. (2000) State-dependent spawn- and Aquatic Sciences 50, 781–8. ing migration in Norwegian spring-spawning herring. Parker, G.A. and Maynard Smith, J. (1990) Optimality Journal of Fish Biology 56, 138–62. models in evolutionary ecology. Nature 348, 27–33. Stöcker, S. (1999) Models of tuna stock formation. Math- Phipps, M.J. (1992) From local to global: The lessons ematical Biosciences 156, 167–90. from cellular automata. In: D.L. DeAngelis and L.J. Strand, E., Huse, G. and Giske, J. (2002) Artificial Gross (eds) Individual-based Models and Appro- evolution of life history strategies and behavior. The aches in Ecology. New York: Chapman & Hall, pp. American Naturalist 159(6). 165–87. Sundby, S. and Fossum, P. (1990) Feeding conditions of Pitcher, T.J. and Parrish, J.K. (1993) Functions of Arcto-Norwegian cod larvae compared with the schooling behaviour in teleosts. In: T.J. Pitcher (ed.) Rothschild and Osborn theory on small-scale turbu- The Behaviour of Teleost Fishes, 2nd edn. London: lence and plankton contact rates. Journal of Plankton Chapman & Hall, pp. 364–439. Research 12, 1153–62. Railsback, S.F. (2001) Concepts from complex adaptive Trebitz, A.S. (1991) Timing of spawning in largemouth systems as a framework for individual based model- bass: implications of an individual-based model. Eco- ling. Ecological Modelling 139, 47–62. logical Modelling 59, 203–27. Railsback, S.F., Lamberson, R.H., Harvey, B.C. and Duffy, Tuljapurkar, S. and Caswell, H. (eds) (1997) Structured W.E. (1999) Movement rules for individual-based Population Models in Marine, Terrestrial and Fresh- models of stream fish. Ecological Modelling 123, water Systems. New York: Chapman & Hall. 73–89. Tyler, J.A. and Rose, K.A. (1994) Individual variability Reynolds, C.W. (1987) Flocks, herds and schools: A dis- and spatial heterogeneity in fish population models. tributed behavioural model. Computer Graphics 21, Reviews in Fish Biology and Fisheries 4, 91–123. 25–34. Tyler, J.A. and Rose, K.A. (1997) Effects of individual Roff, D.A. (1992) The Evolution of Life Histories. New habitat selection in a heterogeneous environment on York: Chapman & Hall. fish cohort survivorship: a modelling analysis. Journal Romey, W.L. (1996) Individual differences make a differ- of Animal Ecology 66, 122–36. ence in the trajectories of simulated fish schools. Eco- Uchman´ski, J. and Grimm, V. (1996) Individual-based logical Modelling 92, 65–77. modelling in ecology: what makes the difference? Rose, K.A., Cowan, J.H., Clark, M.E. and Houde, E.D. Trends in Ecology and Evolution 11, 437–40. (1999) An individual-based model of bay anchovy Vabø, R. and Nøttestad, L. (1997) An individual based population dynamics in the mesohaline region of model of fish school reactions: predicting antipredator Chesapeake Bay. Marine Ecology Progress Series 185, behaviour as observed in nature. Fisheries Oceanogra- 113–32. phy 6, 155–71. Rose, K.A., Cristensen, S.W. and DeAngelis, D.L. (1993) van Winkle, W., Rose, K.A., Winemiller, K.O., Individual-based modelling of populations with high DeAngelis, D.L., Christensen, S.W., Otto, R.G. and mortality: A new method based on following a fixed Shuter, B.J. (1993) Linking life history theory, environ- number of model individuals. Ecological Modelling mental setting and individual-based modelling to 68, 273–92. compare responses of different fish species to environ- Rummelhart, D.E., Hinton, G.E. and Williams, R.J. mental change. Transactions of the American Fish- (1986) Learning representations by back propagating eries Society 122, 459–66. errors. Nature 323, 533–6. Walter, E.E., Scandol, J.P. and Healey, M.C. (1997) A Saila,S.B.andShappy,R.A.(1969) Random movement and reappraisal of the ocean migration patterns of Fraser orientation in salmon migration. Journal du Conseil River sockeye salmon (Onchorhynchus nerka) by International pour l’Exploration de la Mer 28, 153–66. individual-based modelling. Canadian Journal of Salvanes, A.G.V. and Hart, P.J.B. (2000) Is individual vari- Fisheries and Aquatic Sciences 54, 847–58. ation in competitive ability in juvenile cod related to Werner, E.E. and Gilliam, J.F. (1984) The ontogenetic haemoglobin genotype? Sarsia 85, 265–74. niche and species interactions in size-structured popu- 248 Chapter 11

lations. Annual Reviews in Ecology and Systematics Georges Bank larval cod and haddock. Deep Sea 15, 393–425. Research II 43, 1793–822. Werner, F.E., Perry, R.I., Lough, R.G. and Naimie, C.E. Williams, G.C. (1966) Adaptation and natural selection. (1996) Trophodynamic and advective influences on Princeton, NJ: Princeton University Press. 12 The Economics of Fisheries

RÖGNVALDUR HANNESSON

12.1 INTRODUCTION 12.2 THE SURPLUS PRODUCTION MODEL The subject matter of economics is the use of scarce resources to satisfy practically unlimited A natural starting point is the surplus production demands. This is particularly true of the subdisci- model, where the rate of growth of a fish stock in pline of fisheries economics. The productivity of excess of what is needed to compensate for natural wild fish stocks is limited by nature. What makes deaths, or surplus growth (G) for short, depends on fisheries economics particularly relevant is that the size of the stock (S) (Schnute and Richards, the rules governing the fishing industry often fail Chapter 6, this volume; Sparre and Hart, Chapter to take due account of nature’s limited product- 13, this volume, where biomass is called B). This ivity. Economics offers some guidance for what the is, needless to say, a great simplification. The appropriate rules are. usefulness of this model is that it represents the The subdiscipline of fisheries biology is a limited productivity of nature in a simple and good deal older than fisheries economics. Two transparent way and allows one to demonstrate, classic papers on fisheries economics were pub- simply and clearly, some key concepts and lished in the 1950s by two Canadian economists, processes. In reality the growth of fish stocks de- Gordon (1954) and Scott (1955) but there were pends on a number of factors, most of which vary some contributions before that time (see Smith, over time subject to environmental fluctuations. Chapter 4, this volume). The subject ‘took off’ in This has important economic implications which the late 1960s and the 1970s with contributions will be discussed later. Schaefer (1957) used the by Anderson (1973, 1976), Gould (1972), Plourde surplus production model in applied analyses, and (1970, 1971), Smith (1968, 1969), and in particular it is sometimes named after him. by Clark (1973a, 1976, 1985). Textbooks dealing Figure 12.1 shows three possible surplus growth with fisheries economics are Anderson (1986), curves. One is based on the logistic equation, an- Cunningham et al. (1985), Hannesson (1993), and other on a modified logistic equation with a Clark (1976, 1985). Of these, Clark’s books are critical threshold level of viability, and a third the most mathematically advanced. On the is based on yet another modification of the logistic development of the subject, see Scott (1979). equation, which shows depensatory growth but Classic biological reference works, useful for the no threshold of viability. These differences in economist, are Beverton and Holt (1957), shape have implications for the shape of the sus- Ricker (1975) and Hilborn and Walten (1992). tainable yield curve, as will be discussed below, and by Schnute and Richards (Chapter 6, this volume). 250 Chapter 12

(a) 0.06 Logistic growth curve The surplus growth curve shows how much it is

) 0.05 possible to fish sustainably. Any quantity between G 0.04 zero and maximum surplus growth is sustainable; 0.03 if the stock happens to be in equilibrium at some # # 0.02 level S , we can fish the amount G(S ) per unit of 0.01 time indefinitely, and the stock would remain at # 0 S , because what we are taking away corresponds –0.01 exactly to the surplus growth. The question is, how much should we take? Should we take the –0.02 maximum surplus growth or something less? If we –0.03 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 take less, we see that we could take that quantity Population size (S) from two different stock levels. Should we take it from a ‘small’ stock or a ‘large’ stock? (b) 0.05 Threshold level of viability

)0.04 Population growth rate ( G 0.03 12.3 FISHING EFFORT 0.02 AND FISH YIELD 0.01 The first step towards answering these questions is 0 to investigate the relationship between the activi- –0.01 ties of the fishing fleet and the amount of fish it –0.02 catches. This relationship is in fact highly com- –0.03 plex, depending on the technology used and the re- 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 action of a fish stock to continued exploitation and Population size (S) depletion (Misund et al., Chapter 2, this volume). Fishery biologists early on invented the concept (c) 0.030 Depensatory growth of fishing effort. The purpose was to find an indica- 0.025 tor of the abundance of fish stocks, a quantity not ) Population growth rate (

G 0.020 easily observed. Fishing effort is a measure of the 0.015 activity of the fishing fleet directly aimed at catch- ing fish. It can be rigorously defined as the mortali- 0.010 ty that the fleet causes in a fish stock of a given size 0.005 and distribution (the importance of these latter 0 two qualifications will become clear presently). –0.005 Examples of practically measuring effort are hours Population growth rate ( of trawling (eventually corrected for differences in –0.010 vessel size), number of hooks lying in the water for –0.015 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 a certain number of hours, or days fishing. By defin- Population size (S) ition, fish are caught at a rate equal to the product of instantaneous fishing mortality (F) and stock Fig. 12.1 Three examples of a surplus growth function. size: (a) Logistic: G = aS(1 - S/K), with a = 0.5 and K = 1; (b) logistic with a threshold value of viability (A): Y = FS (12.1) G = a(S - A)(1 - S/K), where A = 0.1 and a and K as before; (c) depensatory: G = aSa(1 - S/K), with a=2 and a and K as before. G = rate of growth; S: size of stock; K: maximum where Y is the rate at which fish are removed from stock size (carrying capacity of the environment). the stock. If fishing mortality is directly propor- tional to fishing effort (f), we have The Economics of Fisheries 251

F = qf (12.2) migrate nearly as widely as the stocks are depleted. But reality departs to a greater or lesser degree from where q is a constant (sometimes called availabil- these stylized examples. There are indications ity or the catchability coefficient). Combining that the northern cod stock of Newfoundland (12.1) and (12.2) we get became more concentrated as it was depleted (Hutchings and Myers 1994). Hence the catch per Yf= qS. (12.3) unit of effort (CPUE) did not fall as rapidly as the stock abundance, which led Canadian fisheries This means that the catch per unit of effort which biologists astray and delayed the necessary cut- is measured over a ‘short’ period of time, since Y is back in fishing. the rate at which fish are removed, is proportional We shall in the following stick to these two to the size of the fish stock being exploited and can polar cases, as they are easy to analyse mathemati- be used as an index of its size. cally and to depict graphically. We shall also stick The problem with this is that the postulated re- to the definition of effort given previously but note lationship between fishing effort and fishing mor- that it is likely to be inadequate for economic pur- tality depends critically on the assumption that poses. For that we would need a measure of effort the fish are always evenly distributed over a given that takes into account all activities that give rise area, or that the relative distribution of effort and to costs. Such a measure would be more compre- fish is always the same in a given area. This is not hensive than ‘fishing’ effort as defined above. In always the case, and perhaps only exceptionally so. addition to catching fish, the activities of a fishing If the assumption holds, the density of the fish is fleet involve steaming to and from the fishing directly related to the size of the fish stock; twice ground, searching for concentrations of fish, and as many fish would mean twice as many fish per handling of the catch and gear. If these activities square kilometre, and twice as many fish would be are always proportional to ‘fishing’ effort there likely to be dragged up per hour of trawling or to would not be any problem, but that is not likely; bite the hooks that lie in the water overnight. But consider, for example, the difference between day- consider a different scenario, one where the area trip boats that always return at night irrespective over which the fish are distributed shrinks in pro- of whether they have filled up the hold or not and portion to the size of the stock. If the size of the boats that store the catch on board and do not re- stock shrinks by one half, the area where the fish turn until the hold has been filled. Because of the are will also shrink by one half. The density of fish obvious problems of generalizing about this we will remain the same as before, in that part of the shall leave the matter at that and use the term area where they are located, and if the fishermen ‘fishing effort’ in the sense already defined as a know where to find them, as modern technology measure of the activity of the fleet and the one that increasingly allows them to do, the catch per gives rise to costs. In any case it would not be easy trawl-hour or hook-night would be the same as to do without the biologist’s notion of fishing ef- before, and it would tell us nothing about the size fort in any applied work, because one would need of the stock. to relate fish production to the size of the exploited In all probability we have identified two polar stock and the activity of the fishing fleet. cases between which reality is likely to lie. The We thus end up with the following two ‘polar’ area over which fish are distributed is likely to relationships between catch and effort: shrink somewhat as the stock is depleted, but not in proportion to the depletion. It appears that Y = qfS (12.4a) bottom-dwelling fish like cod (Gadus morhua L.) do not contract a great deal as a consequence of de- Y = kf. (12.4b) pletion, while surface-dwelling stocks that travel in shoals, like herring (Clupea harengus L.), do not Equation (12.4a) is the case where a stock is always 252 Chapter 12 evenly distributed over a given area and where a (a) 0.06 Logistic growth change in the catch per unit of effort directly re- 0.05 flects a change in the size of the stock. Equation (12.4b) is the case where the area over which the 0.04 stock is distributed is proportional to the stock 0.03 size. In that case the density of fish is always con- Yield stant, and so is the catch per unit of effort, which in 0.02 equation (12.4b) is equal to k. In order to take into 0.01 account the intermediate cases, some authors Y AfSb 0 have used the functional form = , where 0 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20 0 £b£1. Another way to look at this is to regard Effort the availability coefficient q as being dependent on b-1 the stock size; i.e. q = AS . If the density of the (b) 0.045 Threshold level of viability stock is always constant, b=0, and the avail- 0.040 ability would be inversely related to the stock size. 0.035 For attempts to estimate b for cod and herring, 0.030 see Hannesson (1983) and Bjørndal (1987, 1988). 0.250

Empirical evidence of negligible sensitivity of the Yield 0.020 catch per unit of effort to the stock size can be 0.015 0.010 found in Ulltang (1980) and Butterworth (1981). 0.005 0 0 0.02 0.04 0.06 0.08 0.1 12.4 SUSTAINABLE YIELD Effort

Depensatory growth We may now combine equation (12.4) with the (c) 0.025 growth function to derive a relationship between sustainable yield, defined as the catch that is equal 0.020 to the surplus growth, and fishing effort. Three sustainable yield curves are shown in Fig. 12.2, 0.015 corresponding to the three growth curves depicted Yield in Fig. 12.1 and the production relationship in 0.010 equation (12.4a) (equation (12.4b) simply produces a straight line with a slope k, up to a level deter- 0.005 mined by the maximum surplus growth). The 0 logistic growth equation produces a sustainable 0 0.01 0.02 0.03 0.04 yield curve of a similar shape, while the curve with Effort the critical threshold level and the one with depen- satory growth give rise to sustainable yield curves Fig. 12.2 Sustainable yield curves for the growth that look like loops. Such loop-like curves may functions shown in Fig. 12.1 and the catch function also arise if there are ‘diminishing returns’ to the Y = qfS, with q = 1. stock, i.e. if instead of (12.4a) we have the catch equation Y = AfSb, with b<1. (See also Schnute and Richards, Chapter 6, this volume; Sparre and Hart, where the access to the fish stock is open to all and Chapter 13, this volume.) free of charge. Assume that the price of fish (P) is Before answering the question what level of constant and independent of the volume of land- sustainable yield we should go for, let us consider ings, and that the cost per unit of fishing effort (C) what would happen in an unregulated fishery is also constant. Assume further that all fishing The Economics of Fisheries 253 boats are identical in every respect. The value of Cf ('high' C) sustainable yield per unit of effort will then be PY/f and identical for all boats. Sustainable yield will be obtained when the biological system is in equilib- PY rium, which can only happen when the fishing Cf ('low' C) fleet is also in equilibrium, that is, when there is no investment in new boats. What, then, determines the investment in new boats? Presumably, people Revenue, cost will invest in new boats if the value of the catch per unit of effort (or catch per boat) is greater than the cost per unit of effort (the cost of each boat). Hence, the system will not be in equilibrium unless the Effort value of the catch per unit of effort is equal to the cost per unit of effort: Fig. 12.3 Two bioeconomic equilibria with open access. Equilibrium occurs where the cost line crosses the catch PY f= C. (12.5) value curve. A lower cost leads to greater effort (f), a smaller stock, and (in this case) a lower sustainable yield. The sustainable catch value is derived using a constant Combining this with equation (12.4) we get price and the logistic surplus growth function and assuming that the catch per unit of effort is proportional PqS = C (12.6a) to the stock.

Pk > C or Pk = C or Pk < C. (12.6b)

In case (12.6a) there will always be some value of S duces the industry to apply greater effort, which in for which an equilibrium exists, provided there is the end results in a smaller sustainable yield. This no threshold value of viability for the stock. Figure foreshadows the conclusion that open access may 12.3 illustrates two possible equilibria for this not be the most appropriate ‘rule of the game’ for case. The equilibrium occurs where the sustain- the fishing industry. able catch value (left-hand side of (12.6a) after The existence of an equilibrium value of S is, multiplying by f) is equal to the total cost (right- however, no guarantee that it will in fact be at- hand side of (12.6a) after multiplying by f). The tained. The trajectory towards the equilibrium figure uses the logistic surplus growth equation may have the form of a spiral, and it may spin away G = aS(1 - S/K) (see also Schnute and Richards, from the equilibrium rather than approach it. This Chapter 6, this volume). Setting this equal to the is what happens for equilibria in the lower part of catch (qfS) gives S = K(1 - qf/a) and a sustainable the loop-shaped sustainable yield curves in Fig. catch value of PqfK(1 - qf/a). The two equilibria 12.2. Furthermore, if there is a critical threshold obtain for a ‘low’ versus ‘high’ cost of effort, respec- value below which the stock cannot reproduce, the tively (high versus low price of fish would give the equilibrium will not even exist for a high enough same kind of comparison). For example, if the cost price of fish or a low enough cost of effort. Consider of effort fell, then effort would increase, ‘disturb’ again equation (12.6a), which determines the equi- the equilibrium and drive the stock down to a new librium level of the stock. From this we see clearly and lower equilibrium level. that a lower cost or a higher price implies a lower Note that the equilibrium may be such that ex- equilibrium value of S. Figure 12.4 shows two equi- cessive effort is used, in the sense that a given sus- librium values of S, one for a low cost (or high price) tainable yield is taken with greater effort than and one for a high cost (or low price). The figure necessary. This happens in our example as a result also shows the direction of movement of effort and of technological progress: a lower cost of effort in- stock when out of equilibrium, and the spiral 254 Chapter 12

capacity over a short time-span. The herring EE2 EE1 stocks may possibly have been saved by a morato- rium that was put in place around 1970. This story alerts us to the fact that nothing except a high cost of catching the last viable animal protects wild animal stocks that are hunted freely from being )

f hunted to extinction, as has in fact happened for certain stocks where the hunting conditions come

Effort ( close to being described by equation (12.6b). A paper by Smith (1975) is a fascinating exposition on why this may explain the extinction of the indi- genous American horse (Equus caballus) several thousands of years ago. Other examples, such as the Kiwi bird (Apteryx spp.) and the American buffalo (Bison bison), come to mind. Both these Stock size (S) animals were an easy prey to the early settlers in New Zealand and on the North American prairies, Fig. 12.4 Bioeconomic equilibrium. The downward- respectively. sloping line shows all combinations of effort (f) and A special form of open access obtains when the stock (S) where the yield is sustainable; i.e. where Y = G, total catch is controlled but everyone is free to par- and Y = qfS, and G is given by the logistic equation. The EE-lines show the stock level compatible with economic ticipate in the fishery. This typically leads to short- er and shorter fishing seasons, and little or nothing equilibrium (equation 12.6a), where EE1 implies a low price or a high cost, and EE2 a high price or a low cost. is accomplished from an economic point of view. The arrows show movement of variables when out For a discussion of this, see Homans and Wilen of equilibrium, and the dashed line shows a possible (1997). pattern towards equilibrium.

12.5 OPTIMAL movement towards the equilibrium point (for EXPLOITATION unstable equilibria we would spiral away). Paths towards equilibrium for real-world fisheries have The stage has now been set for considering eco- been analysed by Bjørndal and Conrad (1987) and nomically optimal exploitation of fish stocks. Conrad (1989). When deriving the optimum sustainable yield of The risk of extinction as a result of fishing is fish stocks it is imperative to keep in mind that most dramatic, however, if equation (12.6b) holds. fishing is only one of the activities that contribute In this case no equilibrium exists except by coinci- to our welfare. It need not make sense, for example, dence; the value of the catch per unit of effort will to achieve the maximum sustainable yield, as we always be the same, because the density of the fish might be forsaking too much of other goods for that will always be the same, until the last fish has been purpose. The fundamental criterion for having taken. This case may come close to describing the achieved optimum sustainable yield is that the situation for the Atlantic herring stocks which last unit of effort expended in the fishery should nearly collapsed in the late 1960s, due to an in- produce the same value as it would do if used in the creased pressure caused by the introduction of the best alternative way. If the marginal unit of effort power block which made it possible to haul purse produces a greater value in the fishery than else- seines mechanically instead of by hand. This in where, then obviously it makes sense to increase turn made it possible to use much bigger seines and effort in the fishery, and vice versa. boats, and led to an enormous increase in fishing This implies that effort can be used for purposes The Economics of Fisheries 255 other than fishing. In the long term this is certainly condition for optimality is that the value produced true; people employed on fishing boats could be by the last unit of fishing effort applied should be employed somewhere else, and capital to be in- equal to the cost of that unit. In the language of vested in fishing boats could be invested in some- differential calculus this can be expressed as thing else. In the short term, however, reality may look a bit different. Fishing boats are typically not PdYdf()= C. (12.7) very useful for other purposes and can only be con- verted to other uses at some cost, and fishermen Figure 12.5 illustrates the solution and compares it may need retraining to be employed in other indus- with what obtains under open access, for the case tries. This implies that the short-term opportunity described by equation (12.4a) and the logistic sur- cost of effort (i.e. the value that the effort could pro- plus growth equation. Clearly it is not worthwhile duce in other industries) may be lower than indi- to take the maximum sustainable yield. Less effort cated by the wage rate or the capital cost on the should be used than needed for that purpose, and firms’ books. But in the long run the capital cost the fish stock should be kept at a higher level than and the wage rate are likely to reflect the value that corresponds to the maximum sustainable yield. labour and capital are able to produce in their best Note also that the optimum effort is less and the alternative application, and we shall in our analy- equilibrium stock greater than what would result sis below assume that the cost of effort measures under open access. The yield is not necessarily the value that effort could produce elsewhere in greater, however, than under open access, unless the economy. the effort under open access is sufficiently greater What, then, is the value that effort produces in than needed to take the maximum sustainable the fishery? Below we shall take this as being syn- yield. Loop-shaped yield curves, such as shown in onymous with the value of the catch. This pre- Fig. 12.2, would produce revenue functions of a supposes that fish are valuable only in so far as they similar shape. This case is not very interesting, contribute to our material well-being as a source of however, as equilibria occurring in the lower part food or raw materials and that this is correctly of the curve are unstable. measured by the market price. This is likely to be In the case of equation (12.4b) it would always true or nearly true in a great many cases; fish are be optimal to take the maximum sustainable sought at considerable inconvenience, and even yield. This is so because the value of the catch per risk of life and limb, for the purpose of selling unit of effort and the cost per unit of effort are both them in the marketplace. There are, however, constant in this case; the value of the catch per unit other cases where fishing as such has value, such as of effort does not decline as the stock is depleted, in recreational fishing (Cowx, Chapter 17, this due to the constant density of the stock. volume). Clearly, in such cases it would not be Note that the optimum sustainable yield was enough to measure the value produced by fishing derived by setting the value of the marginal sus- effort by the value of the fish being caught, and the tainable yield equal to the unit cost of effort. If the time spent on this activity would not even be a unit cost of effort rises with effort, the marginal cost. Lastly fish stocks as such may have value, sustainable revenue should be equal to the margin- for reasons of maintaining biodiversity or for al cost of effort (i.e. the cost of the last unit). If the viewing as wildlife (Reynolds et al., Chapter 15, unit cost of effort is constant it is equal to the mar- this volume). ginal cost. We would have arrived at the same With these caveats we set out to derive the opti- condition if we had maximized the profit in the mum sustainable yield and the associated fishing fishery: i.e. the difference between revenue and effort. In doing so we regard the fishery as being cost [PY(f) - Cf]. We did not do so, in order to em- managed by a single owner, a social planner trying phasize that maximization of profit is not an obvi- to maximize the total value of production derived ously legitimate social goal. It would make perfect from all resources at disposal in the economy. The sense for somebody who owned the resource, but it 256 Chapter 12

(a) would not be a primary goal from a social point of 0.06 view. What does make sense from a social point of view is to maximize the value produced by the re- 0.05 sources at society’s disposal. This occurs when the last unit of any productive resource produces the 0.04 same value irrespective of where it is used. This implies, however, that the profit in the fishery is 0.03 being maximized. This ‘profit’ is a bit special, as it is due to the limited productivity of the fish stock 0.02 and can be seen as a cost of using that productivity. Total revenue In the economic jargon this goes under the name of 0.01 Total cost resource rent, or fishing rent, due to its analogy with land rent. 0 0 0.03 0.06 0.09 0.12 0.15 0.18 Like land rent, the fishery rent is a residual that remains after all factors of production (labour, Effort capital and other inputs) have been paid. The rent (b) reflects the differences in productivity between 1.0 different ‘quality’ categories of a resource. Inner city plots yield a higher rent than plots in the sub- urbs, because of a better location for business. Fer- 0.5 tile agricultural land can be rented out or sold at a higher price than poorer land. The profits realized by extracting oil from a well in the sands of Arabia 0 are higher than the profits that can be extracted from underneath the North Sea. And the profit ob- Average revenue tainable from fishing cod in the Norwegian Sea is –0.5 Marginal revenue higher than fishing the less coveted saithe in the Cost per unit effort same area, provided both stocks are properly man- –1.0 aged. When fisheries are controlled by some fish- 0 0.03 0.06 0.09 0.12 0.15 0.18 ing rights scheme, such as individual transferable Effort quotas or transferable boat licences, the resource rent becomes capitalized in the form of a value of a Fig. 12.5 Equilibrium in an open access fishery versus fish quota, or a value of a fishing boat with a licence optimum sustainable yield. (a) The curve shows the in excess of what the boat is worth for the purposes sustainable revenue (value of the sustainable yield) at a constant price, and the line shows the total cost of effort. of fishing only. If such rights are handed out free of Equilibrium with open access occurs where total cost is charge they end up as windfall gains for those who equal to total revenue, whereas optimum effort occurs got these rights initially and who later sell them in where the difference between sustainable revenue and the marketplace. Note, however, that it is possible total cost is greatest (i.e., where the dotted line is to reduce or perhaps eliminate these gains by im- tangential to the sustainable revenue curve). (b) The posing a fee on quotas or licences, or by selling downward sloping lines show the sustainable revenue per unit of effort (thin line) and the marginal sustainable them or renting them out. revenue (P(dY/dZ), thick line). Equilibrium with open Maximization of profit in the fishery would, access occurs where the sustainable revenue per unit of however, under certain circumstances, lead us effort is equal to the cost per unit of effort (the horizontal astray. Suppose that the price of fish from a particu- dotted line) whereas the optimal effort is where the lar stock depends on the landings from that stock marginal sustainable revenue is equal to the cost per unit only because, for example, the fish is sold in a local of effort. market and there is no other fish that competes The Economics of Fisheries 257 with it. We can write the sustainable profit, alias late into profits being obtained by fishermen who rent (V), in the fishery as are better skilled than others, or have better equip- ment than others. Such ‘skill rents’ or ‘equipment Vf()= PYf[]() Yf()- Cf. (12.8) rents’ are often a conspicuous fact of life in real- world fisheries even under open access. The impli- Maximizing this requires that cations of a volume-dependent price have been analysed by Anderson (1973), and the fishing rents dV df=+[] P() dP dY Y() dY df-= C 0. (12.9) by Copes (1972).

This would not be the socially optimal solution. The value produced by the last unit of effort is 12.6 TIME DISCOUNTING P(dY/df), as this is what the consumers of fish are willing to pay for the fish caught by the last unit of Up to now we have been concerned with sustain- effort. This should be equal to C, the cost of effort. able yield. If we put an equal emphasis on what The term dP/dY is negative, because the price will happens in the short and the long run this is all that be lower the more fish is being landed. If the fishery matters, but if we value any given benefit we get in were controlled by a single firm, or an industrial the future less than if we get it now, it is not enough association, it would presumably be aware that just to look at sustainable yields. it would be eroding its own market by selling The systematic ‘devaluation’ of effects that more, assuming that all fish must be sold at the occur in the future is called discounting. The im- same price, and it would take this into account plications of time discounting have been analysed when deciding how much to sell. This would be by Clark (1973a, 1973b, 1976), Clark et al. (1973), an example of the exercise of monopoly power, and Clark and Munro (1975). The ethical underpin- which is not in the social interest. This mo- nings of time discounting are often called into nopoly power should not be confused with sole question, as it would seem to amount to a system- ownership of scarce resources like fish stocks; al- atic discrimination against future generations. though it makes no sense socially to artificially There is, however, another argument in favour of limit the supply of something just to get a higher discounting. If it is possible to invest profitably in price in the market, it makes perfect sense to limit the economy, we should require that all invest- fishing to what the productivity of nature can sup- ment opportunities yield the same return at the port, with due account taken of the cost of fishing. margin, or else only invest in such opportunities as This is what sole ownership, or privatized use yield the highest return. By making profitable rights, of scarce resources would attain, in contrast investments we do in fact leave a richer world to to the overexploitation occurring under open our descendants. Discounting the future stream of access. benefits from any investment at the same rate of Now that we have relaxed the assumption of a return as we can get in the best alternative oppor- constant price of fish we might as well ask: what tunity is a method for ascertaining whether that would this mean for our previous analysis of open investment is in fact worth while. access versus optimum exploitation? The answer To explain this, suppose we can invest our is: not a great deal; open access would still result in money in the bank so that it will yield r ¥ 100% overexploitation. Similarly, relaxing the assump- interest every year. Note that this is not a purely tion that the cost per unit of effort is constant does financial phenomenon; the rate of interest in the not cause any fundamental change; what happens banks may be expected to reflect rates of return on is that all the fishing rent would not be absorbed by ‘real’ investments: that is investment in produc- unnecessary costs; all units, except the last one, tive capacity. The reason that the banks can charge would obtain some profit over and above their op- a certain rate of interest is that somebody is pre- portunity cost. In everyday life this would trans- pared to borrow the money and pay it back with 258 Chapter 12 interest, financed out of profits that he expects to model. The term e-rt is the analogue of the dis- make on the investment. count factor (1 + r)-t in continuous time. If interest If we deposit the amount K in the bank we accrues n times per year, one unit of money would would have K(1 + r)T at the end of T years. Suppose grow to (1 + r/n)nt over t years. Then consider the instead that we invest the amount K at time 0 in a expression [1 + 1/(n/r)](n/r)rt. The expression [1 + 1 project that will provide an income I net of operat- /(n/r)](n/r) approaches e as n approaches infinity. ing cost every year for T years, after which our in- Hence the present value of exploiting the stock in vestment is worthless. At the end of the T years we perpetuity is would have I[1 + (1 + r) + ...(1 + r)T-1] = I[(1 + r)T - 1]/r, assuming that we continuously invest our in- • PG() S PV= PG() S e-rt dt = (12.11) come from the project in the bank as it accrues. Ú r Hence, if the project is worth while, we must have 0

-T where r is the discount rate, which equals the rate IrrK11-+() > . (12.10) [] of return we can earn on an alternative investment project. The immediate gain of increasing fishing The left-hand side of this is the so-called present by the amount S will be equal to P S (note that value of the income stream I over T years, dis- -D - D changing the amount fished will cause an opposite counted at the rate r. Hence the criterion for a change in the stock size). This will change the profitable investment is that the income stream present value of all future catches by from a project, discounted at a rate of interest equal to the return on the alternative investment, be at least equal to the initial outlay for the project. DDPV= [] P() dG dS r S. (12.12) The relevance of this is that we can regard any fish that we do not catch immediately as an invest- If DS represents a departure from an optimal stock ment. Why should we leave it in the sea? Because a level, to be maintained in perpetuity, the sum of fish left uncaught contributes to the growth of the these two changes must be zero: i.e. the short-term stock, through individual growth and through re- gain must be cancelled by the long-term loss. production. If fish did not grow, or did not grow fast Hence, PDS - [P(dG/dS)/r]DS = 0, or enough, it would make no sense to leave them in the sea. Hence, if we exploit a fish stock optimally, dG dS= r. (12.13) the return on a fish we leave in the sea must be equal to the return we can get on catching that fish, The solution is illustrated in Fig. 12.6. We see that selling it in the marketplace and investing the the optimum equilibrium stock is in fact smaller money we get for it at the highest return we can than that which corresponds to maximum sus- obtain. tainable yield. In other words, a positive discount What about fish stocks that do not grow fast rate implies that some biological overexploitation enough to satisfy the required rate of return? On would be optimal. The reason for this is that dis- the basis of the above reasoning such stocks should counting of the future makes it worth while to be fished out and converted to other forms of capi- incur a permanent loss for the sake of a temporary tal that are more productive. Many people un- gain. Even if the absolute value of a permanent loss doubtedly find such a recommendation offensive, is infinite, its present value when we discount the but implicit in that attitude is that fish stocks are future is finite; the positive discount rate turns the valuable for other purposes than their surplus infinite series of losses into one that converges to a production, such as for preserving biodiversity or finite value. for tourism. Taking fishing costs into account modifies this Consider now a stock that is optimally exploit- conclusion, provided the catch per unit of effort ed. For easier exposition we use a continuous time depends on the size of the exploited stock. In The Economics of Fisheries 259

dG/dS < 0 in the optimal solution. This would cer- Slope = r tainly be true in the absence of discounting (r = 0). Thus biological overfishing need not be optimal, even if the future is discounted. But discounting of )

G the future reduces the optimal standing stock; the higher the discount rate (r) is, the greater is dG/dS, and the smaller is the optimal standing stock (see Fig. 12.6). Growth rate ( The implications of discounting could be dra- matic. If max dG/dS < r and the unit cost of landed fish is not stock-dependent (X is constant), the implication is that the stock should be fished to o S Smsy Stock size extinction; investing in the stock simply would (S) not yield a high enough rate of return to be worth while. The implication is that such stocks should Fig. 12.6 Optimum stock level (So) with a positive be ‘mined’, like minerals or oil deposits, which discount rate when the catch per unit of effort does after all are resources with too low a rate of growth not depend on the stock level. Smsy is the level giving maximum sustainable yield. (zero or, for oil, negligible) to make exploitation based on surplus growth interesting. There are a number of slow-growing fish and whale stocks that might be in this category. Orange roughy (Hoplostethus atlanticus) is a slow-growing fish that case it is attractive to fish from a large stock which matures when it is 30 years old and lives to rather than a small one, in order to keep down be 60 to a 100 years old if left unfished. Yields were the cost per unit of fish caught. Let the cost per high in this fishery when it started and the stocks unit of fish caught be denoted by X(S). The im- were mined, but the sustainable yields have turned mediate gain from increasing the amount fished out to be much lower (see, for example, Batstone by -DS is -(P - X)DS. The present value of future and Sharp 1999). Given that the stocks of such fishing is now species are sufficiently valuable as such it would not, of course, be optimal to mine them to extinc- • []PXSGS- () () PV=-Ú[] P X() S G() S e-rt dt = . tion, but the point is that if the exploitation of such 0 r stocks were a matter to be decided by the industry (12.14) in its own interest, or by a sole owner, the invest- ment aspect would be likely to prevail. The indus- The change in the present value resulting from try or a sole owner would not attach much value to changing the stock by DS is now the stocks as such; these values stem from ethical considerations like preserving species for their DDPV=-{}[]() P X() dG dS- () dX dS G r S. own sake, which are not likely to loom large in the (12.15) profit and loss accounts of private individuals or firms. Letting immediate gains be cancelled by perma- Above we have looked at optimal equilibria. nent losses now gives Another question is what the adjustment path towards the equilibrium will be like if, say, the rdGdSdXdSGPXS- ()+ ()[]- ()= 0. (12.16) fishery starts from a situation with overexploita- tion. The optimal approach path can be shown to Since dX/dS < 0 (fishing from a larger stock reduces depend on the discount rate and to what extent the cost per unit of fish landed), it is possible that capital, measured in terms of production equip- 260 Chapter 12 ment, is ‘malleable’. The reader is referred to a clas- d22 R dY= 2() dP dY+ () d 22 P dY Y. (12.20) sic paper by Clark et al. (1979). Because larger landings normally imply a lower price, dP/dY < 0. We are not assured that d2R/ 12.7 FLUCTUATIONS: dY 2 < 0 but it is quite likely. SHOULD CATCHES As an example, consider the demand function BE STABILIZED? P = AY -b. (12.21) It was mentioned above that the growth of fish stocks is influenced by fluctuations in the marine The revenue function is environment (Myers, Chapter 6, Volume 1). These 1-b fluctuations give rise to variations in the size of R = AY , (12.22) fish stocks, even if they are exploited at a rate con- sistent with long-term sustainability. The fluctua- and its derivatives are tions in stock abundance in turn give rise to -b fluctuations in catches. Such fluctuations are usu- dR dY=-()1, b AY (12.23) ally considered an inconvenience by the industry, 22--b 1 as fish buyers usually prefer even and secure deliv- d R dY=- b()1 - b AY . (12.24) eries. A review of stochastic models in fisheries is 2 2 provided by Andersen and Sutinen (1984). If 0 < b < 1, dR/dY > 0 and d R/dY < 0 and we have a One way of responding to the problems concave revenue function. The opposite is true if caused by fluctuations is to stabilize the catch b > 1; then we have a convex revenue function, (Hannesson and Steinshamn 1991). Would this be ER > R(EY), and fluctuating catches would in fact desirable, from an economic point of view? Con- yield a higher revenue on the average than a stable sider, first, the revenues from fishing. These are catch would do. In this latter case the price is high- ly sensitive to the quantity being sold, and the rev- RY()= PYY(). (12.17) enue becomes greater as the quantity becomes smaller, because the price ‘goes through the roof’. Now suppose that the fishery is managed by a TAC This case is known as inelastic demand (the elas- (total allowable catch) regime where the TAC is set ticity of demand, -dlogY/dlogP, is equal to 1/b). equal to some fraction of the stock available in It may be noted that stabilizing the catch would each period. Suppose, further, that the stock is sub- imply stabilizing the effort as well if the density of ject to random fluctuations. Then the TAC will fish is always constant (equation 12.4b). With a fluctuate as a consequence. constant unit cost of effort the cost would simply Would a stable catch equal to the expected be proportional to the catch and irrelevant for catch (EY) with a fluctuating TAC provide a larger whether or not the catch should be stabilized. revenue? It would, if Things turn out differently if the cost of landed fish depends on the size of the exploited stock. Let RY()EE> RY() (12.18) us proceed on the basis of equation (12.4a), where the catch per unit of effort is proportional to the where ER is the expected revenue under the fluctu- size of the stock. We normalize effort such that Y = ating TACs. This holds if the revenue function is fS. Setting the TAC equal to a certain fraction of concave, with dR/dY > 0, at least for ‘low’ values of the stock amounts to fixing the effort at the level f* Y, and d2R/dY 2 < 0. This is quite possible. From which, because of the normalization, is equal to equation (12.17): the desired fraction to be caught from the stock. Under this catch policy the total cost of fishing dR dY=+ P() dP dY Y, (12.19) will in fact be constant and equal to Cf*. If, on the The Economics of Fisheries 261 other hand, the catch is held stable at a level equal option economically, we turn to the case of fluctu- to the expected catch under the said TAC policy ating catches. (Y* = EY), we would have to vary effort according to equation (12.4a): 12.8 OPTIMUM FLEET fYS= *. (12.25) CAPACITY FOR FLUCTUATING STOCKS The expected effort would be If fish stocks fluctuate in part for random reasons EEfY= *.()1 S (12.26) and the management regime sets a TAC that is somehow related to stock abundance, such as a Using Y* = f*ES, we can write this as given fraction of the stock or everything in excess of some target escapement, the optimum fleet EEEffS= *.()1 S (12.27) capacity will not depend solely on the determinis- tic stock–growth relationship element that influ- Since 1/S is a convex function of S, E(1/S) > 1/ES, so ences the development of the stock; the nature of that ESE(1/S) > 1. Hence, Ef > f*, which means that the random fluctuations will also be a determinant stabilizing the catch will be more costly than let- of optimum capacity. In this case optimal manage- ting it vary with the stock. What the stabilization ment involves determining the capacity of the policy amounts to is telling the industry to fish fishing fleet and its use at any particular time. In with great intensity when the stock is small and ‘bad’ years the capacity of even an optimal fleet the catch per unit of effort is low but to hold back will be too great and its activities must be some- when the stock is plentiful and the catch per unit how restricted. These decisions are interrelated: of effort is high. Hence, even if a stable catch were that is, the use of an existing fleet, or the optimum attractive for marketing and for the processing TAC in any particular year, depends on the optimal industry, it would be less attractive, and perhaps size of the fleet, provided the stock size is not en- decidedly unattractive, for the catching industry, tirely random but depends as well on a determinis- if the catch per unit of effort depends on the abun- tic stock–growth relationship. This latter issue is dance of the stock. Hence there is no clear econom- discussed in Hannesson (1993) but here we shall, ic argument for stabilizing catches; on the revenue for simplicity, ignore the deterministic stock– side there are likely to be arguments pulling in growth relationship and assume that the growth of that direction but on the cost side the argument is the stock is entirely random. likely to point the other way. Another simplification we shall make is that Note, finally, that stabilizing the catch is likely the catch per unit of effort is constant and that ef- to be a risky option. It amounts to fishing inten- fort is simply proportional to the number of identi- sively when the stock is low and less intensively cal vessels in the fleet. The value of the catch each when it is plentiful. This may jeopardize the year will then be growth of the stock, particularly if there is a threshold level of viability or if low stock levels are R= min() PQ , PaK (12.28) somehow inimical to growth as occurs when there is depensation. Some stocks fluctuate so wildly where Q is the total allowable catch, K is the size of that the TAC is set equal to zero in some years; the fishing fleet, measured as capital investment, capelin (Mallotus villosus) is an example. Stabiliz- and a is the amount one unit of money invested ing the catch at a biologically safe level might in a in the fleet can catch if the fleet is used to its full case like that imply the ridiculously low level of capacity. Here P is the price net of operating costs, zero. Given that stabilizing the catch is risky or which would include fuel, ice, fishing gear, and impossible, and in any case not a clearly superior other expendable equipment. 262 Chapter 12

As explained above, we can find the optimal ) Q ( policy by maximizing the present value of the rent F 1 in the fishery. We make the further simplification 1–F(aK o) = (r + δ)/Pa that we start without any fleet and that the boats we build have no alternative use. The best policy will then be to invest in an optimal fleet at time F(Q) zero and maintain it forever by setting aside the fraction dK of revenues to cover depreciation of Cumulative distribution o the fleet (d=1/T, T being the lifetime of a fishing Qmin aK Qmax vessel). The present value of rents is then Total allowable catch (Q)

0 • Fig. 12.7 Optimal fleet capacity (K ) for a total ERt - d K VK =- +Â t . (12.29) allowable catch (Q) that fluctuates randomly between t=1 ()1+ r Qmin and Qmax. F(Q) is the cumulative distribution function. The linear form of F implies that the If the probability distribution of the stock (and the probability density is constant. The arrows show the TAC) is time-invariant, ER will be the same year effect of a higher price, lower capital costs, or after year. Then the sum in the above expression technological progress (higher a). will be a convergent geometric series and we get

VKERKr=- +() -d . (12.30) Figure 12.7 shows the optimal solution, with We get a more intuitive interpretation of this if we arrows indicating how a lower capital cost (r +d), a multiply on both sides by r (the maximum of V will higher price net of operating cost (P), or technologi- occur for the same value of K if we multiply V by a cal progress (greater a) affects the solution. We see scalar). Then we get that a lower capital cost, a higher price net of oper- ating cost, or technological progress all increase VERrK*.=-+()d (12.31) the optimum fleet capacity, as indeed we would expect. We also see that only exceptionally would What we are in fact maximizing is the rent per year, it be economically sensible to invest in a fleet that calculated as the revenue net of operating cost less is large enough to take the maximum TAC (the Q- capital cost. The capital cost consists of two ele- value for which F(Q) = 1). This would involve in- ments, the depreciation of capital, d, and the oppor- vestment in boats that most of the time are not tunity cost, r, of tying up capital in a fishing fleet needed, contributing to cost all the time but to rather than investing in something else giving the revenue only infrequently. Only if the cost of capi- return r. The condition for maximum rent is tal is exceptionally low would it make sense to have a fleet that could take even the largest TAC. EdRdK()=+d r . (12.32) Note also how a change in the probability distri- bution affects the optimal fleet capacity. Suppose What, then, is E(dR/dK)? From equation (12.28) we it is realized that the minimum total allowable see that dR/dK > 0 only when aK < Q, i.e. when the catch could be much lower than Qmin in Fig. 12.7 total allowable catch is greater than the capacity of but that the probability density of Q remains uni- the fleet; only then will another vessel make any form between Qmax and the new Qmin. This would contribution to the total catch. Define F(Q) as the rotate the line in Fig. 12.7 clockwise (it is fixed at probability that the total allowable catch is less the point 1, Qmax) and thus reduce the optimum than or equal to Q. Then we have fleet capacity. In this section we have regarded the fleet as []1- FaK()=+() rd Pa. (12.33) being managed by a single owner (the social plan- The Economics of Fisheries 263 ner again). Although this approach is useful to de- optimum level of effort, or by rotating the average rive optimality conditions, from the point of view sustainable revenue line downwards until it inter- of the whole economy, this is not how the real- sects the cost per unit of effort line above the opti- world fishing fleets are run. A pertinent question mum level of effort. These movements would be is: will individual decision makers achieve the accomplished by putting a tax equal to tf on effort, optimality? Under open access they certainly or ty on the landings of fish, such that will not, but in a market-driven management oo regime such as individual transferable quotas P()1- ty Y() f f= P() dY df fo (12.34) (ITQs), to be discussed below, they might. There is a presumption, however, that they will in fact or overinvest when the crew is remunerated by a share in the catch, but ITQs are likely nevertheless CtPdYdf()1+ f = ()fo (12.35) to lead to much less overinvestment than open access. For details of this analysis, see Hannesson where fo is the optimal level of effort. Either tax (2000). would confiscate all rents in the fishery and deter the fishing firms from investing in a bigger fishing fleet than needed to take the optimum sustainable 12.9 MANAGEMENT yield. METHODS This solution is neat in theory, but even within the confines of the deterministic model it is not as Based on the deterministic, surplus production straightforward to apply as it may seem. Fishing model it has been concluded that all we need to do effort is produced by a number of different inputs: to manage fisheries for maximum economic bene- manpower, capital, fuel etc. All of these would fit is to tax fishing effort or the landings of fish suf- have to be taxed proportionately in order not to dis- ficiently to eliminate all incentives for economic turb the cost-minimizing combination of inputs. overfishing. The argument is perfectly valid with- Taxing landings would appear more straightfor- in the framework of the said model, provided ward, but the optimal tax is determined by the that the catch per unit of effort increases with the growth characteristics of each stock, so that if stock level. For fluctuating stocks there is a case many stocks are being fished by the same fleet, the for control by landing fees, provided there is no landings from each stock would have to be taxed economic uncertainty and the catch per unit of ef- at a different rate. Finally, there is the case where fort depends on the stock size in a stable and pre- the density of fish is always constant (equation dictable way. Landing fees versus output controls 12.4b). Here the value of the catch per unit of (fish quotas) have been discussed by Weitzman effort is constant. An optimal tax on landings or (2002). on effort would equalize the revenue per unit of We look, first, at the deterministic biomass- effort and the cost per unit of effort for all levels of growth model. Consider sustainable yield and the effort but would fail to identify any particular level production relationship in equation (12.4a), which of effort. gives rise to Fig. 12.5. As explained above, the But the most difficult problems occur when we open-access equilibrium occurs where the average enter the world of fluctuating stocks and TACs. sustainable revenue is equal to the cost per unit Note that these fluctuations typically occur on a of effort, while optimal exploitation requires that much shorter time-scale than the lifetime of fish- the cost per unit of effort be equal to the marginal ing vessels. We need, therefore, to contract and sustainable revenue. We may in fact achieve the expand fishing effort over the lifetime of a single optimum solution by moving the line showing boat, perhaps over several cycles. This is not easily cost per unit of effort upwards until it intersects accomplished by changing tax rates. Such changes the marginal sustainable revenue line above the more often than not are time-consuming, unless 264 Chapter 12 the management authorities get a mandate to do so ing away at sea of undersized fish or fish that are swiftly and substantially as the circumstances not covered by a quota. might require. But even if that mandate existed, Boat licences are related to effort controls. This the management authority would not know pre- method can be used for controlling the capacity of cisely how the industry would react to a change in the fishing fleet, but it has several limitations. the tax. It might even react in a direction opposite Fishing capacity is a multidimensional variable, to what the authority would expect and opposite to and it is difficult to control all of the variables opti- what would happen in the long run. A fisherman mally. In order to be effective a boat licence has to who has to care for a mortgage or two, a wife, a specify the size and design of a boat in some detail. child, a dog and a car might, in the face of a reduced Naval architects have been notoriously inventive income due to a higher tax, decide to increase his in circumventing such regulations, packing an im- effort in order to make ends meet in the short run, pressive amount of fishing capacity into a hull that even if in the long run he could not renew his boat meets length or tonnage requirements. In some if the high tax were to prevail. Trial and error cases such designs have been alleged to reduce the would presumably teach the authorities how fish- seaworthiness of fishing vessels. Fishing licences ermen respond to the tax changes, but that learn- therefore are an imprecise method of controlling ing process might need a longer time-scale than fleet capacity and one that causes unnecessary tolerable to cope with short-term fluctuations in costs. Individual transferable catch quotas (ITQs) fish stocks. would appear to do so much more effectively by af- It is difficult, therefore, to see how one can do fecting the incentives to invest, provided they can without some direct control of the fishing activ- be effectively implemented. ity, in order to cope with random fluctuations in catches. Two major modes of control are available, Individual transferable quotas a control of fleet capacity and fishing effort, and a control of the catch through catch quotas. Catch quotas have frequently been used for keep- ing the total catch from a stock within desired Effort controls limits. As soon as fisheries started to be controlled by TACs it became evident that this caused new Effort controls only indirectly achieve the short- problems in fisheries where the capacity of the term objective of keeping the catch of fish within fishing fleet exceeded that which was necessary to some set limit. These types of control may be very take the permitted catch. Often a fierce competi- imprecise in this regard; the relationship between tion developed for getting the largest possible effort and catch is seldom precisely known and is share of the TAC. In many cases management re- likely to depend on the size of the stock and climat- sponded by dividing the TAC among the boats in ic and oceanographic conditions. Fishermen may the fishery. get around effort controls if all elements of effort From this, ITQs evolved. When the TAC was are not included: a limited number of fishing days much less than the catch capacity of the boats, it may be made ineffective by using more gear, for was clear that cost savings could be achieved by example. The only advantage effort controls allowing people to trade quotas among themselves would seem to have compared to a direct control of rather than having each and every one go out and the catch is that they may be much easier to moni- fish his perhaps very small quota. In that way the tor; fishing boats can be seen and counted, their owners of active boats could make ends meet, and trips and even type of activity can be monitored, those who were eligible to participate in the fish- nowadays by satellite tracking, and the gear and ery but chose not to got a share of the pie by renting other equipment they use can be inspected. Con- out their quota. Making the quota allocation valid trolling landings is, on the other hand, often costly for a long time eliminates the incentives to invest or nearly impossible, to say nothing of the throw- purely for the purpose of getting a share of the rents The Economics of Fisheries 265 in the fishery, an activity which in the end is self- extent distort the system of incentives and entice defeating; in the long run the rents get absorbed by boat owners to invest more than is desirable from unnecessary costs, as has already been explained. an overall perspective (Hannesson 2000). With a quota allocation that is secure for a period at There is a large and still increasing literature on least as long as the lifetime of a fishing boat the ITQs. The idea may be traced back to Christy quota holder can predict his future catches, using (1973, 1975). Arnason (1995) and Batstone and the best available biological evidence. He would Sharp (1999) contain descriptions of ITQs in Ice- have no incentive to invest in a fishing boat which land and New Zealand. Boyce (1992, 1996) consid- is larger or better equipped than needed to take the ers bycatch and other problems under ITQs. A expected future catches, and if a bigger boat is more much-quoted critique is Copes (1986). Among cost effective he would be able to acquire an addi- other contributions are Grafton (1996), Hannesson tional quota allocation by buying it from some- (1996, 1997a) and Weninger (1998). body else. ITQs can be determined either as fixed tonnages or shares of the TAC. Using fixed tonnages for 12.10 INTERNATIONAL stocks where the TAC varies from year to year ISSUES makes it necessary for the fisheries manager to buy and sell quotas, depending on whether the TAC is Management by TACs or other methods affecting above or below the total amount of quotas al- catches and fish stocks is meaningful only if the located. If the quota tonnage is set low enough the country in question can exercise effective control manager would be selling quota more often than over the stocks. The establishment of the Exclu- buying and would be making money on these trans- sive Economic Zone (EEZ) in the 1970s was critical actions (Hannesson 1989). This would amount to a in this respect (for an authoritative text, see United special tax on the fishing industry. Tonnage quotas Nations 1983; and Christy and Scott 1965, for a dis- transfer the risk associated with fluctuating stocks cussion preceding the revolution in the law of the from the industry to the fisheries manager, usually sea in the 1970s). In some cases fish stocks became the government. That risk can indeed be greater enclosed by the EEZ of a single sovereign state, than the government is prepared to bear. Fixed ton- while in other cases stocks migrate between the nage quotas were initially tried in New Zealand, zones of two or more sovereign states, so no single but when it became clear that the yield potential of state can exercise effective control over them. the orange roughy stocks had been overestimated Nevertheless, in many cases the states concerned the government backtracked and redefined the have agreed on effective controls over shared quotas as shares of the TAC (Batstone and Sharp stocks, such as setting an overall TAC and dividing 1999). This is the system in use in most of the ITQ it among themselves. Having accomplished this, systems in the world today. With share quotas the each country can apply measures such as ITQs to industry has to bear the risk associated with fluctu- its own share of the TAC, for the purpose of maxi- ating TACs, but quota holders can still make ratio- mizing its economic benefit from its share of the nal predictions of the catches they may expect to fishery. Norway has, for example, concluded such get on the basis of their quota, using the best evi- agreements with the Soviet Union, subsequently dence available about fluctuations in the stocks taken over by Russia, and the European Union. they fish and the criteria on the basis of which the The member countries of the European Union TAC is set. Hence boat owners do not have any have agreed among themselves on the division of obvious incentives to overinvest under a share the North Sea stocks, but fisheries policy is one of quota system. The so-called share system, by the common policies of the Union. which labour employed on fishing boats is remu- Still there are many stocks that spend a part nerated with a share in the catch value and not of their life history outside any EEZ, and some through a fixed wage rate, may, however, to some stocks are mainly or even wholly confined to the 266 Chapter 12 high seas. Management of such stocks, and of other players. If the N-1 players wait, the best the shared stocks, must be by voluntary consent by the remaining player can do is not to wait, as S > S/N. parties concerned, as these are sovereign states. It And if the N-1 players do not wait, the best the re- is particularly difficult to achieve such agreements maining player can do is to do likewise, because for stocks on the high seas, not least because the otherwise he would get nothing. number of interested parties is indeterminate. This certainly makes for a pessimistic outlook Game theory seems an appropriate tool for with respect to the possibility of achieving a mutu- analysing the question of whether nations will ally advantageous solution. The situation is not, come to agreement on fisheries issues. There however, well represented by this framework. already exists a voluminous body of literature on Fishing is not a one-shot game; fish stocks are re- this which it is impossible to review here. Impor- newable, and fishing is an activity that is repeated tant references are Munro (1979), Levhari and year after year. What if we look at a strategy which, Mirman (1980), Vislie (1987), Kaitala and Munro loosely speaking, means that ‘I’ll be nice to you as (1997), and Hannesson (1997b). Here a simple long as you’ll be nice to me, but if you’re nasty to model which at least gives a flavour of the issues me I’ll be nasty to you for ever after.’ In this setting will be presented. this would mean that a representative player Suppose a stock can be fished either at an imma- would wait as long as all the rest do likewise, but if ture or a mature stage. There is no stock–growth one player does not wait for the fish to grow, the relationship to be taken into account; each period a others will do the same forever. The present value stock S emerges and can be fully depleted without of the strategy of playing cooperatively forever is jeopardizing future recruitment, but if the ex- 2 ploitation is delayed by one period, the stock will PVSgc =+()11()+ rNSg++() 11()+ rN+ ... grow to S(1 + g), where g is the rate of growth. Sup- =+SgrN()1 , (12.36) pose N identical countries have access to the stock. If they all wait for one period and allow the stock to while the present value of not playing co- grow, each would get S(1 + g)/(1 + r)N. The discount operatively is rate appears because such waiting is an invest- 2 ment, as already explained. If they fish the stock PVn =+ S S()11 + r N++ S() r N+=+... S S rN . immediately, each would get S/N, but if all except (12.38) one decide to wait they would wait in vain; the one who does not play along could take it all. Cooperation is profitable if PVc > PVn, which This gives rise to the following payoff matrix, implies which shows the return to one player. The lines show the strategy of the player in question, while gN> r, (12.39) the columns show the strategy of the other players, all assumed to adopt the same strategy. which can be interpreted as saying that each player’s share in the rate of growth of the stock has to be greater than the rate of interest. Clearly, for Wait Don’t wait reasonable rates of growth and discount rates, the Wait S/N 0 number of participants does not have to be very Don’t wait S S(1 g)/(1 r)N + + high to make voluntary agreements unlikely. If the cost per unit of fish landed depends on the stock Provided g > r, it would be better for all to wait. The the likelihood of agreement increases, but decreas- so-called Nash equilibrium in this game does not es again if the costs differ among participants result in that solution, however. A Nash equili- (Hannesson 1995). This game can also be seen as brium is a situation in which no player can gain by yet another example of how discounting the future changing his strategy, given the strategy of the makes it worth while to trade off permanent losses The Economics of Fisheries 267 against a short-term gain; if r is zero equation graphic entries just cited and the further leads that (12.39) will always be satisfied, as long as N is one would find in these references. finite. The most recent attempt to come to grips with uncontrolled fishing on the high seas is the UN REFERENCES Agreement on Straddling Stocks and Highly Migratory Stocks, concluded in 1995. One source Andersen, P. and Sutinen, J.G. (1984) Stochastic bioeco- for the text of this agreement is FAO (1995), and a nomics: a review of basic methods and results. Marine review of the status of this agreement is provided Resource Economics 1, 117–36. by Munro (2000). According to this agreement, the Anderson, L.G. (1973) Optimum economic yield of a fish- ery given a variable price of output. Journal of the Fish- authority to manage such stocks rests with region- eries Research Board of Canada 30, 509–18. al management organizations. Any country with a Anderson, L.G. (1976) The relationship between firm and ‘real interest’ (not further defined) has a right to fishery in common property fisheries. Land Econom- become a member of such an organization, and all ics 52, 180–91. countries, even those who are not members, are Anderson, L.G. (1986) The Economics of Fisheries Man- obliged to abide by the decisions of such organiza- agement. Baltimore, Md.: Johns Hopkins University Press. tions. The above analysis indicates that these orga- Anderson, L.G. (1993) Toward a complete economic the- nizations will not be likely to succeed unless they ory of the utilization and management of recreational manage to limit the number of interested parties fisheries. Journal of Environmental Economics and and the quantity they are allowed to fish. The Management 24, 272–95. latter is not possible unless there is an effective Arnason, R. (1995) The Icelandic Fisheries. Oxford: Fish- system in place to punish violators. At the present ing News Books. time this is up to the state whose flag the offending Arnason, R., Hannesson, R. and Schrank, W.E. (2000) Management costs in fisheries. Marine Policy 24, vessel is flying, with some rights of other interest- 233–43. ed states to inspect suspected offenders and to take Batstone, C.J. and Sharp, B.M.H. (1999) New Zealand’s action if the flag state does not do so. quota management system: the first ten years. Marine Policy 23, 177–90. Beverton, R.J.H. and Holt, S.J. (1957) On the dynamics of 12.11 CONCLUSIONS exploited fish populations. Fishery Investigations, series II, vol. 19. London: HMSO. Bjørndal, T. (1987) Production economics and optimal It is impossible, within the limited space at our stock size in a North Atlantic fishery. Scandinavian disposal, to cover all issues that arise in fisheries Journal of Economics 89, 145–64. economics. The most important ones have, I hope, Bjørndal, T. (1988) The optimal management of North been covered but some still remain. Of those re- Sea herring. Journal of Environmental Economics and maining the most important ones are manage- Management 15, 9–29. ment and enforcement costs (see Schrank et al. Bjørndal, T. and Conrad, J.M. (1987) The dynamics of an open access fishery. Canadian Journal of Economics 2002; Sutinen and Andersen 1985), the recreation- 20, 74–85. al use of fisheries (see McConnell and Sutinen Boyce, J.R. (1992) Individual transferable quotas and pro- 1979; Anderson 1993), and marine protected areas duction externalities in a fishery. Natural Resource (see Lauck et al. 1998; Hannesson 1998; Sanchirico Modeling 6, 385–408. and Wilen 1999). The last topic has aroused quite a Boyce, J.R. (1996) An economic analysis of the fisheries bit of interest over the last few years, but this ap- bycatch problem. Journal of Environmental Econom- proach is unlikely to provide a panacea for all fish- ics and Management 31, 314–36. Butterworth, D.S. (1981) The value of catch-statistics eries management problems and is certainly no based management techniques for heavily fished substitute for ITQs or other methods that would pelagic stocks with special reference to the recent de- get us close to optimal management. On these sub- cline of the Southwest African pilchard stock. In: K.B. jects we refer the interested reader to the biblio- Haley (ed.) Applied Operations Research in Fishing. 268 Chapter 12

NATO Conference, series II, vol. 10. London: Plenum ry and practice. Reviews in Fish Biology and Fisheries Press, pp. 441–64. 6, 5–20. Christy, F.T. (1973) Fisherman Catch Quotas. University Hannesson, R. (1983) The bioeconomic production func- of Rhode Island, Law of the Sea Institute, Occasional tion in fisheries: a theoretical and empirical analysis. Paper 19. Canadian Journal of Fisheries and Aquatic Sciences Christy, F.T. (1975) Property rights in the world ocean. 40, 968–82. Natural Resources Journal 15, 695–712. Hannesson, R. (1989) Catch quotas and the variability Christy, F.T. and Scott, A.D. (1965) The Common Wealth of allowable catch. In: P. Neher, R. Arnason and N. in Ocean Fisheries: Resources for the Future. Balti- Mollett (eds) Rights Based Fishing. Proceedings of more, Md.: Johns Hopkins Press. the NATO Advanced Workshop on the Scientific Clark, C.W. (1973a) The economics of overexploitation. Foundations for Rights Based Fishing, Reykjavik, Science 181, 630–4. 27 June–1 July, 1988, Wageningen: Kluwer, pp. Clark, C.W. (1973b) Profit maximization and the extinc- 459–65. tion of animal species. Journal of Political Economy Hannesson, R. (1993) Bioeconomic Analysis of Fisheries. 81, 950–61. Oxford: Fishing News Books. Clark, C.W. (1976) Mathematical Bioeconomics. New Hannesson, R. (1995) Fishing on the high seas: coopera- York: Wiley. tion or competition? Marine Policy 19, 371–7. Clark, C.W. (1985) Bioeconomic Modelling and Fisheries Hannesson, R. (1996) Fisheries Mismanagement: The Management. New York: Wiley. Case of the North Atlantic Cod. Oxford: Fishing News Clark, C.W. and Munro, G. (1975) The economics of fish- Books. ing and modern capital theory: a simplified approach. Hannesson, R. (1997a) The political economy of ITQs. In: Journal of Environmental Economics and Manage- E.K. Pikitch, D.H. Huppert and M.P. Sissenwine (eds) ment 2, 92–106. Global Trends: Fisheries Management. Proceedings of Clark, C.W., Edwards, G. and Friedlaender, M. (1973) a Symposium held at Seattle, Washington, 14–16 June, Beverton–Holt model of a commercial fishery: optimal 1994. Bethesda, Md: American Fisheries Society, pp. dynamics. Journal of the Fisheries Research Board of 237–45. Canada 30, 1629–40. Hannesson, R. (1997b): Fishing as a supergame. Journal of Clark, C.W., Clarke, F.H. and Munro, G.R. (1979) The Environmental Economics and Management 32, optimal exploitation of renewable resource stocks: 309–22. problems of irreversible investment. Econometrica Hannesson, R. (1998) Marine reserves: what would they 47, 25–49. accomplish? Marine Resource Economics 13, 159–70. Conrad, J.M. (1989) Bioeconomics and the bowhead Hannesson, R. (2000) A note on ITQs and optimal invest- whale. Journal of Political Economy 97, 974–91. ment. Journal of Environmental Economics and Man- Copes, P. (1972) Factor rents, sole ownership and the agement 40, 181–8. optimum level of fisheries exploitation. The Man- Hannesson, R. and Steinshamn, S.I. (1991) How to set chester School of Economics and Social Studies 41, catch quotas: constant effort or constant catch? Jour- 145–63. nal of Environmental Economics and Management Copes, P. (1986) A critical review of the individual quo- 20, 71–91. tas: a device in fisheries management. Land Econom- Hilborn, R. and Walters, C.J. (1992) Quantitative fish- ics 62, 278–91. eries stock assessment: choice, dynamics and uncer- Cunningham, S., Dunn, M.R. and Whitemarsh, D. (1985) tainty. New York: Chapman & Hall. Fisheries Economics: An Introduction. London: Homans, F.R. and Wilen, J.E. (1997) A model of regulated Mansell Publishing. open access resource use. Journal of Environmental FAO (1995) Structure and Process of the 1993–1995 Unit- Economics and Management 32, 1–21. ed Nations Conference on Straddling Stocks and Hutchings, J.A. and Myers, R.A. (1994) What can be Highly Migratory Fish Stocks. FAO Fisheries Circular, learned from the collapse of a renewable resource – No. 898. Rome: FAO. Atlantic cod Gadus morhua, of Newfoundland and Gordon, H.S. (1954) The economic theory of a common Labrador. Canadian Journal of Fisheries and Aquatic property resource: the fishery. Journal of Political Sciences 51, 2126–46. Economy 62, 124–42. Kaitala, V. and Munro, G.R. (1997) The conservation and Gould, J.R. (1972) Extinction of a fishery by commerical management of high seas fishery resources under the exploitation: a note. Journal of Political Economy 80, new law of the sea. Natural Resource Modeling 10, 1031–8. 87–108. Grafton, R.Q. (1996) Individual transferable quotas: theo- Lauck, T., Clark, C.W., Mangel, M. and Munro, G.R. The Economics of Fisheries 269

(1998) Implementing the precautionary principle in Scott, A.D. (1955) The fishery: the objectives of sole own- fisheries management through marine reserves. Eco- ership. Journal of Political Economy 63, 116–24. logical Applications 8 (1 supplement): s72–s78. Scott, A.D. (1979) Development of economic theory on Levhari, D. and Mirman, L.J. (1980) The great fish war: an fisheries regulation. Journal of the Fisheries Research example using a dynamic Cournot–Nash solution. Board of Canada 36, 725–41. Bell Journal of Economics 11, 322–34. Smith, V.L. (1968) Economics of production from natural McConnell, K.E. and Sutinen, J.G. (1979) Bioeconomic resources. American Economic Review, 58, 409– models of marine recreational fishing. Journal of 31. Environmental Economics and Management 6, 127– Smith, V.L. (1969) On models of commercial fishing. 39. Journal of Political Economy 77, 181–98. Munro, G.R. (1979) The optimal management of trans- Smith, V.L. (1975) The primitive hunter culture, Pleis- boundary renewable resources. Canadian Journal of tocene extinction, and the rise of agriculture. Journal Economics 12, 355–76. of Political Economy 83, 727–56. Munro, G.R. (2000) The United Nations fish stocks Sutinen, J.G. and Andersen, P. (1985) The economics of agreement of 1995: history and problems of im- fisheries law enforcement. Land Economics 61, plementation. Marine Resource Economics 15, 387–97. 265–80. Ulltang, Ø. (1980) Factors affecting the reaction of pelagic Plourde, C.G. (1970) A simple model of replenishable re- fish stocks to exploitation and requiring a new ap- source exploitation. American Economic Review 60, proach to assessment and management. In: A. Saville 256–66. (ed.) The Assessment and Management of Pelagic Fish Plourde, C.G. (1971) Exploitation of common-property Stocks. Conseil International pour l’Exploration de la replenishable natural resource. Western Economic Mer, Rapports et Procês Verbaux des Réunions 177, Journal 9, 256–66. 489–504. Ricker, W.E. (1975) Computation and Interpretation of United Nations (1983) The Law of the Sea. United Biological Statistics of Fish Populations. Ottawa: De- Nations Convention on the Law of the Sea, with Index partment of the Environment, Fisheries and Marine and Final Act of the United Nations Conference of the Service. Law of the Sea. New York: United Nations. Sanchirico, J.N. and Wilen, J.E. (1999) Bioeconomics of Vislie, J. (1987) On the optimal management of trans- spatial exploitation in a patchy environment. Journal boundary renewable resources: a comment on Munro’s of Environmental Economics and Management 37, paper. Canadian Journal of Economics 20, 870–5. 129–50. Weitzman, M. (2002) Landing fees vs. harvest quotas Schaefer, M.B. (1957) Some aspects of the dynamics of with uncertain fish stocks. Journal of Environmental populations important to the management of the Economics and Management 43, 325–38. commercial marine fisheries. Journal of the Fisheries Weninger, Q. (1998) Assessing efficiency gains from indi- Research Board of Canada 14, 669–81. vidual transferable quotas: an application to the Schrank, W.E., Arnason, R. and Hannesson, R. (eds) mid-Atlantic surf clam and ocean quahog fishery. (2002) The Cost of Fisheries Management. London: American Journal of Agricultural Economics 80, Ashgate. 750–64. 13 Choosing the Best Model for Fisheries Assessment

PER SPARRE AND PAUL J.B. HART

13.1 INTRODUCTION approach here is to provide a suite of examples, which we believe to be representative of con- Fisheries scientists have developed a suite of temporary fisheries models. Throughout we have models for a variety of purposes ranging from fun- tried to follow three general principles: (1) models damental descriptions of fish growth to long-term should be as simple as possible, (2) the complexity predictions of population dynamics. While many of a model should match the objective and the of the chapters in this volume review the basic available data, and (3) models should be judged by concepts, a full description of these could easily their utility rather than by the beauty of the math- have occupied both volumes of this book. For de- ematics, because they are tools, not objectives in tailed reviews of fisheries models we recommend themselves. the excellent books by Hilborn and Walters (1992) Mathematical models are used to distil a com- and Quinn and Deriso (1999), and for a more gen- plicated system into a simpler one that we can read- eral overview, we recommend Jennings et al. ily understand and use for predictions. Our ability (2001). It is also still worth consulting Ricker to model the real world depends on our ability to (1975). Our objective in this chapter is more mod- collect adequate data. Smart models cannot substi- est: to outline the basic concepts behind fisheries tute for bad data. It may take hours to develop a modelling, and review briefly the major modelling mathematical model and months to implement it options that confront the fisheries biologist. We on computer, but it may take decades to collect an hope that this overview will help readers to see adequate time series of quality fisheries data. The how the various methods outlined in Chapters descriptive part of a fisheries research programme, 5–12, Volume 2, fit together, as well as providing perhaps using the techniques of Geographical links with various chapters in both volumes that Information Systems (GIS), (Meaden and Do Chi review the basics of fish biology. 1996), may sometimes be more valuable than a There is a risk in saying a little about models mathematical model, which in the worst case may without saying ‘everything’ about them, because be a poor reflection of the key mechanisms revealed fisheries managers may become tempted to grab by the data. Models may not necessarily enhance an ‘off the shelf’ model without paying enough at- our understanding of a fisheries system. tention to its underlying assumptions, or ensuring that it is applied to an appropriate situation. It is imperative to look at each situation afresh and to 13.2 BASIC CONCEPTS choose models that can be tailored to each situa- tion. It is also important to test or ‘confront’ A general expression for a mathematical model is models with data (Hilborn and Mangel 1997). Our useful for the presentation of guidelines for the Choosing the Best Model for Assessment 271

choice of model (Schnute 1994). A model is usually also that a control vector Zt = (Z1t, Z2t,..., Zrt) rep- framed in the form of a mathematical function, resents r quantities known to influence the popu- which can be written as lation dynamics. Then the two vector equations

FXP:,()ttÆ Y , (13.1) FX()tt-1,, F P= X t (13.5a) where Xt and Yt are variables at time t and P is GX()tt,, Z P= Y t (13.5b) a parameter. What equation (13.1) says is that the function F takes values of Xt and P and maps them encapsulate descriptions of system dynamics and onto values of Yt such that there is a unique value measurement, respectively, where P = (P1, P2,..., of Yt for each pair of values (Xt, P). In a fisheries con- Pk) represents a parameter vector of dimension text Xt could be some state of the fish stock and Yt k that remains constant through time. Thus the the catch. The dynamics of the state of the stock, vector F governs the evolution from time t to t + 1 say its size, is likely to be a function of not only bio- of the m states Xt, and G relates the n observations logical processes such as births and deaths due to Yt to the current system state Xt. As outlined by natural causes but also deaths due to fishing. As Schnute and Richards (2001 and Chapter 6, this fishing is a variable that can be controlled we need volume), the model (13.5a)–(13.5b) describes the to know the relationship between Xt and the fish- progression of Xt through the space of all possible eries-related control variable Zt. So we need a new states, and is therefore called a state space model. function G that will describe the way in which Xt A concrete realization of (13.5a) and (13.5b) is changes in response to Zt-1. This is evaluated at t - the discrete form of the logistic equation 1 because it is assumed that there is a lag between the application of the control and the change in the Ê Bt ˆ BBBttt+1 =+a 1 - - qEtt B (13.6a) state of the fish stock. The new function G is then Ë B• ¯

GZ:.()tt-1 Æ X (13.2) Ct = qEtBt (13.6b)

Equation (13.1) can now be rewritten as where Bt is biomass of fish, Et is fishing effort and Ct is catch, all at time t. a is the intrinsic rate of nat- (13.3) FGZ:,()()tt PÆ Y . ural increase, B• is the carrying capacity and q is the catchability coefficient. In terms of our general

So in general terms we can say that the dynamics of form in (13.5a) and (13.5b), m = n = r = 1, k = 3, Xt = the system is fully accounted for by two functions Bt, Zt = Et, Yt = Ct, and P = (a, B•, q). of X, Z and P, A model may be deterministic or stochastic. The deterministic model yields a one-to-one rela-

FX:,,()tt-1 ZPÆ X t (13.4a) tionship between each X and each Y. With the de- terministic model we assume that there is no

GXZP:,,()ttÆ Y t . (13.4b) process error meaning that the equation is perfect, and we can make a precise prediction of Y. This In most cases of interest F and G and the vari- outcome is unlikely in fisheries science. There is ables Xt, Yt, Zt, and the parameters P, are multival- always some degree of uncertainty when predict- ued and have to be described by vectors. As a result ing the development of individual fish, the size of a a fish population will be characterized in each year fish stock, or the capture rate of a fishing fleet. A t by a vector of state variables Xt = (X1t, X2t,..., stochastic model operates with many Y-values for Xmt) with dimension m. Yt = (Y1t, Y2t,..., Ynt) will each X-value, as will be discussed later. Determin- denote the corresponding vector of n observations istic models are easier to handle than are stochas- obtained annually from this population. Suppose tic models. Also the computations and the data 272 Chapter 13 demands for stochastic modelling are greater than the model, one might as well have used the simpler those for deterministic models. Whether to use single-species model. deterministic or stochastic modelling is an im- The model may be applied to estimate the para- portant choice. meters, or to predict the value of Y for an assumed

The choice of parameter vector, P = (P1, P2,..., value of X. The latter application is applied to an- Pk), and the choice of the relationship between X, swer ‘what-if-then’ questions. For example, what Z, P and Y, here called, F and G, is essentially the will happen to the landings, Y, if the fishing effort, choice of the model. The number of parameters (k) Z, is a certain number of fishing days per year? In may be small, say 2 or 3, of it may be hundreds or terms of the logistic equation of a fishery (13.6), thousands. The number of parameters is an impor- one might ask what happens when Et is increased tant choice. You may choose to cover many details by 10%? in your description of the world, or you may choose The model may be dynamic or static. A dynam- to apply the minimum number of parameters. ic model predicts output for a time series of years, There are advantages and disadvantages of both whereas the static model predicts the average approaches. The model-designer is often not free output. A static model may be said to predict what to choose parameters, but is more or less con- will happen ‘on average’, say, over the next 10 strained by the objectives of the model. It may be years, whereas the dynamic model predicts what mandatory to introduce certain parameters into happens in each individual year. It is usually easier the model to meet certain objectives. Alterna- to handle the static than the dynamic models, be- tively, some parameters may have to be left out, cause there are more details to describe in the dy- because the data available cannot be used to esti- namic model, and consequently more parameters. mate them. This is a typical situation for fisheries An example from fisheries of a dynamic model models, with the number of unknowns greatly ex- is the model of the Pacific halibut (Hippoglossus ceeding the number of observations (Schnute and stenolepis) fishery by Thompson and Bell (1934) Richards 2001). (Shepherd and Pope, Chapter 8, this volume). This It may be possible to choose plausible values for predicted numbers of fish dying over a 13-year parameters that cannot be estimated. The philo- period when fishing mortality increased from 40% sophical question faced is whether it is better to to 60%. The yield-per-recruit model of Beverton ignore a feature of the system description or to and Holt (1957) is an example of a static model, as use the best possible ‘guesstimate’ available. Obvi- is the Ricker stock–recruitment curve (Myers, ously there is no general answer to the question. Chapter 6, Volume 1) which gives the average re- Typical examples of fisheries models with cruitment for a given spawning stock biomass. Sta- few parameters are the surplus production models tic models in fisheries are most often chosen (equations (13.6a) and (13.6b), and Chapter 6, this because data for dynamic models are not available. volume), which do not account for the size or age Often the static models are called ‘Long-term pre- composition of stocks. Models accounting for size diction models’ and the dynamic models are called or age usually have many more parameters. Going either ‘Short-term prediction models’ (that, is 2–3 from single-stock models to multi-stock models years) or ‘Medium-term prediction models’ (10–20 clearly increases the number of parameters. Simi- years) (See also Shepherd and Pope, Chapters 7 and larly, going from a one-fleet and one-area model to 8, this volume). a multi-fleet and multi-areas model increases the number of parameters yet again. The number of parameters may not be proportional to the number 13.3 STOCHASTIC of components, but may be more as there may be MODELLING a need to model interactions between the various components. If interactions such as that between No model in fisheries can predict the exact value fish stocks or fishing fleets are not accounted for in of any variable. To the simple model one should Choosing the Best Model for Assessment 273 ideally add stochastic vectors, d and e, so the stochastic model reads as 6000 FX,, Z P X (13.7a) ()tt-1 +=d t 5000

GX()tt,, Z P+=e Y t . (13.7b) 4000

The stochastic vectors d and e take unpredictable 3000 values from a probability distribution which we Frequency may have some knowledge about. Usually, d and 2000 e are assumed to be normally or log-normally distributed in fisheries models. The stochastic 1000 term accounts for all the elements not accounted 0 for by F(Xt-1, Zt, P) and G(Xt, Zt, P). If the model Y1 Y2 Y3 Y4 Y5 Y6 Y7 Y8 Y9 Y10 Y11 Y12 actually reflects the true relationship between X Output and Y, which is rarely the case in any fisheries model, the stochastic term has a known mean Fig. 13.1 Output from a stochastic simulation in terms value which is usually zero. However, the fisheries of the equation Y +e=F(X, Z, P). models are always incomplete with an unknown bias. the result of process error. A discussion of how to In more detail a stochastic version of (13.6a) and distinguish between the two sources of error is (13.6b) requires three probability distributions to given in Hilborn and Walters (1992). be defined. These are The way in which Yt may vary is shown in (Fig. 13.1). Where stochasticity is accounted for in

PPP()()()XXtt,,-10 Z,P t , YXZ,P ttt ,,, YP ,. (13.7a) and (13.7b) by the vectors of residuals d and (13.8) e, they may have a probability distribution of an as- sumed form or a distribution estimated from time

The third item determines the initial state of the series of observations of the pair (Xt, Yt). system. Any distribution can describe the varia- When using a stochastic model for prediction, tion in Xt, Yt and X0. the standard procedure is to let a computer pro- The stochastic terms, d and e, can be thought of gram repeat the same prediction or simulations as containing two components, the measurement many times, say 1000 times or 10000 times (the error and the process error. Each can be expressed Monte Carlo method; Hilborn and Mangel 1997). as a linear combination of two terms, d=dm +dp and In each simulation the computer program draws e=em +ep. The measurement error (dm, em) is the de- the values of parameters from a random-number viation expected if the model is correct, and the generator. Eventually, the probability distribution process error (dp, ep) is the deviation caused by in- is estimated by the frequency distribution of Y (see sufficiencies in the model (Schnute 1994). In a dy- Fig. 13.1). This means that a stochastic model re- namic model, the process errors will appear in the quires as input the characteristics of the probabil- entire dynamic system, whereas the measurement ity distribution for the parameter estimates. For errors will occur individually in each time step. a general introduction, see for example Hilborn As an example of process error and measurement and Mangel (1997) and Manly (1997). One of the error, we may consider a research survey, which earliest attempts at introducing a stochastic ele- suddenly gives an unexpectedly high density of ment into a fishery model was by Beddington and fish. This may be due to bad survey design, which May (1977). They attached stochastic process vari- would lead to measurement error, or to an unex- ation to the parameter a of the Schaefer surplus pectedly high stock productivity, which would be yield equation (equations (13.6a) and (13.6b), and 274 Chapter 13

Objectives (request(s) from Funding customers) manpower

Model Constraint specifications

Input data yet to Constraint Data already be collected available X' X

Constraint Choice of model(s) Fig. 13.2 Flowchart for selection F(X, Z, P) = Y + ε of model comprising the steps: Data Data Define objectives and identify Feedback Data existing data and data to be collected subject to available Estimate P funding. Then choose the input, Parameters Feedback to X, the control variables, Z, the customer output, Y, the parameters, P, the Test of model(s) relationships between X, Y and P, F(X, Z, P) = Y? the stochastic term, e. Then estimate parameters, P, test the model and eventually apply it. For Apply model further explanation see text.

Schnute and Richards, Chapter 6, this volume) and method from first principles. In all cases the task of showed that the coefficient of variation of yield in- choosing a model can be broken into five steps: creased as effort increased, so making the predic- 1 Choosing the states, X = (X1, X2,..., Xm). tion of yield less and less reliable as the fishery 2 Choosing the control variables Z = (Z1, Z2,..., took more and more fish. Zr). 3 Choosing the output, Y = (Y1, Y2,..., Yn). 4 Choosing or estimating the parameters, P = (P , 13.4 MECHANICS OF 1 P2,..., Pk) and the relationships between X, Z, Y CHOOSING A MODEL and P. 5 Choosing the assumption about the stochastic

The choice of a model will depend on what is re- terms, d=(d1, d2,..., dr) and e=(e1, e2,..., er). quired as output, and will be constrained by data Figure 13.2 illustrates some key elements in the available to estimate parameters. For a manager process of choosing a model. The output, Y, is the requirement is most likely to be how the fish- selected to match some objectives defined either ery will behave under a given measure. Existing by the scientist or perhaps more often by a cus- models tailored to the particular fishery may meet tomer, which is often also the funding body. Also this objective. For scientists doing pure research, the funding and the manpower available for the the model may help to explain the workings of na- creation of the model will place obvious con- ture, and this may often require creation of a new straints on what can be achieved. Choosing the Best Model for Assessment 275

Once the objectives are defined clearly, the next Very often, fisheries scientists are constrained step is to investigate the existing data (X) for use as by managers who want a particular type of out- input to the model. Depending on the funding, one put (Y) from a model. An example might be where may (or may not) consider collection of additional managers want to reduce the amount of discard- data (X¢). For example, to apply an age-structured ing, and to that end, they want to introduce tech- model, age composition data must be available. nical measures. Suppose further that one of the The choice of model should aim at utilizing all measures is to close a certain sea area for fish- relevant and available data. Within each family of ing by certain types of vessels using certain types fisheries model, there are usually a small number of gears with certain mesh sizes. An example is of standard relationships, as listed in Table 13.1. the so-called Plaice-box in the North Sea (Pastoors The differences between models are due mainly to et al. 2000). The output may then be the ‘discard differences in the degree to which input is aggre- ogive’ before the introduction and after the intro- gated. For example, are data given by geographical duction of the closed area. The discard ogive areas, by month or by year, and by fishing fleet? can be parameterized by DL50% and DL75%. To Once the relationships between X, Z, P and Y be able to respond to such a request, a number of have been determined, and this is usually the preconditions must be met. Data must be disaggre- easiest part of the task, the next step is to estimate gated by area, fleet, gear, mesh size, species and the parameters and then test if the model fits the body length. A traditional surplus production data. If the model does not fit the data, it should be model cannot be used, as this type of model does modified. Ideally, one should work simultaneous- not cover many of the features of the problem. ly with a number of alternative models (Hilborn Accordingly, a more complex age-structured and Mangel 1997). In many cases, if the aims of model is required. modelling are known, and the manager is using standard fisheries data, it is possible to use models already available in the literature. 13.5 ESTIMATION A typical example of model choice for the man- OF PARAMETERS agement of a fishery would be a situation where the management agency has data on effort exerted Estimation of the parameters, P, from the obser- by the fleet (XFl) measured as days at sea and land- vations (X,Y), can be made by, for example, mini- 2 ings in weight of each species (YSp). The fishery sci- mization of the so-called modified c criterion (see entist now needs to choose a model which relates for example Sokal and Rohlf 1981): the input variable X to the output variable Y. An example might be YSp = qXFlB where q is the catch- n m 2 ()FX()iv, P- Y iv ability coefficient and B is the biomass vector for c 2 = , ÂÂ FX P (13.9) each species. Alternatively, the job could be done i=1 v=1 ()iv , using a surplus production model (Schnute and

Richards, Chapter 6, this volume). In the service of when v = 1, 2,..., m sets of observation (X1v, X2v, realism, the data on X and Y could be divided by ..., Xmv) are available. There are other equally fleet, area and time in the following way: well-established techniques for parameter esti- mation, such as the maximum-likelihood method.

X1,Fl,p,y,Ar,g = Effort of fleet Fl in time period p Thus, there is a choice of methodology when (e.g. month) of year y in area Ar using the model for parameter estimation, al- using gear g though the results will be almost the same when

Y1,Sp,Sc,Fl,p,y,Ar,g = Landings of species Sp, commer- data fit the model well. The problems emerge cial size category Sc caught by when there is a large variation of observations fleet Fl in time p (e.g. month) of around the values predicted by the model. In this year y in area Ar using gear g case, the choice of estimation model may have 276 Chapter 13 great impact on the estimated values, and some- times in fisheries the methods are so elastic 13.6 THE USE OF and the data so bad, that it becomes almost up to MATHEMATICAL MODELS the researcher to decide the parameter value, by IN FISHERIES manipulating the assumptions behind the esti- mation procedure. It might perhaps be better Models of exploited fish populations have devel- in these sick cases to admit that the estimation oped along several different paths determined by failed, and start the search for either better data the amount of detail included. Early attempts were or better models. constrained by the limited computing power that Together with the estimation of parameter was available just after the Second World War. values, the standard methods of parameter estima- There was heavy dependence on integral calculus tion will provide estimates of the probability dis- with equations that could be solved and there was tribution of the parameter estimates themselves. a drive to find simple functions for growth and These in turn may be used as input for stochastic mortality. These early attempts have left a legacy modelling. that has been used to develop some sophisticated The most common methodology for parameter models, which use methods that were not avail- estimation in fisheries science is represented by able to the pioneers. In this handbook we have the Statistical Analysis System (SAS) computer- acknowledged the legacy by devoting chapters to ized statistical system (e.g. DiIorio 2000). There is surplus yield models (Chapter 6, this volume) and no space here for a deeper review of the mathemati- dynamic pool models (Chapters 7 and 8, this vol- cal features of estimation, but we refer readers ume). In addition other chapters deal with tradi- to the large literature on mathematical model- tional models of growth (Chapter 5, Volume 1) and ling and parameter estimation (see Hilborn and recruitment (Chapter 6, Volume 1). The legacy Walters 1992, and Hilborn and Mangel 1997 for pervades the practice of fisheries management an introduction to the literature). It is strongly but fishery scientists have gone beyond the early recommended always to follow the standards of monolithic models, which tended to ignore differ- generally accepted textbooks or computerized sys- ences between fisheries and fish stocks. Most as- tems for parameter estimation, such as SAS or sumed that the Beverton and Holt dynamic pool the S+ (e.g. Krause and Olson 2000) and to avoid model could be applied to any species, so long as inventing ad hoc estimation methods. The SAS one could estimate the required parameters. In the procedures entitled ‘GLIM’ (General Linear Model) remainder of this section we attempt to show how and ‘MIXED MODELS’ (Littell et al. 1999) appear fishery scientists have developed detailed models to be able to handle almost any estimation problem to cope with the reality not always covered by the encountered in fisheries research. In North Ameri- original formulations. In many ways, the submod- ca some fisheries scientists are using AD Model els discussed are really only filling in the cracks Builder (http://otter-rsch.com/admodel.com, AD that have appeared in the original edifice. There is Model Builder 2001) which is an efficient com- definitely room for fresh thinking as described in mercial package designed for non-linear estima- the various chapters which deal with and develop tion. The AD Model Builder is recommended for the legacy. estimation when models are large, complicated In this section it is helpful to make a distinction and non-linear. The AD Model Builder makes between ‘full’ population models that allow fore- it easy to code almost any model and to investigate casting of yields and stock dynamics under various its statistical properties. The Builder is also re- fishing scenarios, and ‘submodels’ or ‘ingredients’ commended for estimation when large, non-linear that go into these. For example, many models in- models are based on many observations and have corporate body growth rates, and this information many parameters (see also Pitcher, Chapter 9, this is either input directly from data, or represented by volume). an equation, which represents the data. This equa- Choosing the Best Model for Assessment 277 tion enters as a submodel. In many cases these sub- where p is an index of ‘predator type’ as is done models have been introduced to deal with detail in multispecies VPA (Andersen and Ursin 1977; that the earlier models did not account for. There Shepherd and Pope Chapter 7, this volume). M0 are no hard-and-fast rules here, because, for exam- is the natural mortality not caused by predation, ple, an equation that describes relationships be- such as disease, spawning stress and old age and tween stock size and recruitment might serve as NP is the number of predators. Mp may be mod- an end in itself when one wants to predict optimal elled as suggested by Andersen and Ursin (1977): exploitation rates of semelparous species. Alterna- tively it might be a component of a full statistical NRSp p pr, p Mp = (13.13) catch-at-age model. But for presentation purposes, jP= we adopt this distinction. ÂNwSjjjp, j=1

13.6.1 Simple models of fish where Sj,p is a measure of the suitability of prey j as stock dynamics food for predator p and Rp is the food ration for predator p. The multispecies VPA was an exten- Most fisheries models are based on differential sion of the single-species VPA, which aimed at the or difference equations, here exemplified by the estimation of Mp from the investigation of stom- model of the death process which is central to age- ach contents of predators (Sparre 1991; Shepherd structured models of exploited populations: and Pope, Chapter 8, Volume 2).

dNt =-ZNtt or when integrated dt 13.6.2 Models of fishing NNt =-0 exp() Zt , (13.10) Like natural mortality, fishing mortality may be divided into fleet components: where N(t) is the number of survivors from a cohort at time t and Z is the coefficient of total mortal- NFs ity. In fisheries, Z is usually assumed to remain FF= Â Fl , (13.14) constant during a standard time interval such as Fl=1 a month, a quarter or a year. Using a time step of length Dt, we get a difference equation, where Z is where Fl is index of fleet and NF is the number assumed to remain constant from time t to time of fleets. A fleet is a group of fairly homogenous t +Dt: vessels with respect to size, fishing operations and fishing grounds.

DDNNNttttt=-+D = N[]1 - exp() - Zt t . (13.11) FFl may be further modelled, for example by:

It is important to notice that Z refers to the en- FFl,St,a = EFlQSt,FlSFl,St,a, (13.15) tire stock. If the stock is divided into fractions ac- cording to geographical areas, there is need for where St is an index of a stock and a is an index of redefining Z when it becomes area-specific. This age groups. Q is the catchability of fleet Fl catching will be discussed below in connection with ‘areas stock St. Here the F is given the index St to empha- and migration’. size that almost all fisheries are mixed fisheries The total mortality, Z, is composed of fishing catching more than one species, a fact that is often mortality F, and natural mortality M, which again not accounted for in routine stock assessments can be further divided, for example: and certainly not in fisheries where the manage- ment system is still relatively primitive. NP The gear selection (S) by trawls can be modelled MM=+0 Â Mp , (13.12) p=1 by the logistic curve: 278 Chapter 13

N 1 L50% SFl,, St a = ;lns1 = ()3 ;YCwFl= Â Fl, a a , (13.20) 1++exp()ss12LSt, a LL75%%- 50 a=0

s1 s2 =- , (13.16) where N is the number of age groups and w is the L50% a mean body weight at age a. This is the basic form of the dynamic pool model, derived originally in con- where LSt,a is the mean body length of age group a (see MacLennan 1992 for a review of gear selec- tinuous form as an integral equation, by Beverton and Holt (1957) (see Shepherd and Pope, Chapter 8, tivity). L50% and L75% are the lengths at which 50% and 75% of the fish entering the gear is retained, this volume for further details). respectively. The logistic curve takes values The value of the catch can be estimated by ex- between zero and one, starting with zero and tending the model: approaching one as L increases (e.g. Hoydal et al. N 1982). For some other gears like a gill-net a dome- VCwFl= Â Fl,, a aPr Fl a , (13.21) shaped curve may better reflect the selectivity of a=0 the gear (Hovgård and Lassen 2000). where Pr is the price per unit weight. Bioeconomic A simple relationship between mesh size (m) in, modelling is discussed in detail by Hannesson for example, the cod-end of a trawl and the reten- (Chapter 12, this volume). tion of fish of a length L is The input vector, Zt, of the general model, Yt = G(X , Z , P) could include the fishing effort, the L = S m . (13.17) t t 50% Fl,St Fl mesh size and the price, all of which are human in- puts to the fisheries system. The yield and the The catchability may be modelled by a fleet and value of the yield may be the output, and the para- stock-specific constant, but it may also be assumed meters are those listed above which are not vari- to be a function of stock biomass and the year, ables: G [Xt, (Effort, Mesh size, Price), P] = (Yield, accounting for the technical development in fish- Value). Fishing mortality may also be considered ing techniques: output, as it is used as a measure for the perfor- mance of the fisheries system. b QQBabyFl, year =+0 (), (13.18) The development of fishery models has made extensive use of the catch equation (13.19). It lies where Q0, b, a and b are constants and y is at the heart of Beverton and Holt’s (1957) yield- year. This is just one among many other math- per-recruit model, but since then it has been used ematical models one can imagine and it assumes extensively in the development of methods to that catchability increases with time in a linear estimate population size and fishing mortality. fashion. Equation (13.19) is the basis of VPA, which is de- The catch can be estimated as follows. Assume scribed in detail in Chapter 7 (this volume) by that the output from the model might be the catch Shepherd and Pope. The equation also lies at the of age group a fish caught by fleet Fl. If we assume root of statistical catch-at-age methods (Hilborn F and Z to remain constant in time period Dt and Walters 1992). These derive an equation, (Baranov 1918), and we skip the stock index, which, in its simplest form, allows the estimation then the ‘catch equation’ gives the catch as: of initial cohort size, fishing mortality and the catchability coefficient from a multiple linear re- FFl, a gression of catch on effort. Doubleday (1976) pro- CFl, a =--NZtaa()1 exp()D . (13.19) Za duced a non-linear equation relating catch to effort but a linear, and therefore easier to fit, version was The yield in weight of the catch accumulated over derived by Paloheimo (1980). His equation for a age groups may also be estimated as single cohort is Choosing the Best Model for Assessment 279

y-1 Ê Cay, ˆ Ê Ey ˆ 13.6.3 Body growth logÁ ˜ = log()Rqya- -+ qÁ Â E k ˜ Ë Ey ¯ Ë kya=- 2 ¯ When estimating yield it is often useful to have a Ê 1ˆ --Ma , (13.22) way of estimating weight at age. This has been re- Ë 2¯ viewed by Jobling (Chapter 5, Volume 1), who pre- sents the von Bertalanffy equation (von Bertalanffy where C is the catch of age a fish in the yth year, a,y 1934), which expresses the relationship between E is the fishing effort in the yth year, R is the y y-a body weight (or length) and age. Recall that length, number of fish recruited at age 0 in year y - a, q is L , is given as follows: catchability and M is natural mortality. The equa- t tion is in the form Kt()- t0 LLt =-•()1 e (13.25) Y = b + b X + b X , (13.23) 0 1 1 2 2 and the length–weight relationship is which is the equation for a multiple regression. In b Wt = aLt , (13.26) this case b0 = log(Ry-aq), b1 = q, and b2 = M. The equation can be fitted using a standard multiple where L and W are length and weight at time t, L regression package which will give estimates of t t • is the asymptotic length at t =•, K is the rate at Rq, q and M, which is unlike VPA and cohort which L approaches L and t is the age at which analysis. There is no need to guess F for any ages t • 0 the fish would be zero length. These equations as- and normal confidence estimates on the parame- sume a one-to-one relationship between age and ters are possible. In this sense these statistical length, which may not be the case. When this is catch-at-age methods are superior to VPA. Even so, true, age–length keys (ALK) can be used (Pitcher, the validity of the method is entirely dependent on Chapter 9, this volume). For example the assumption that equation (13.10) correctly de- scribes the decay of survivors and that the stock ALKia, =£Pr() L i length < L i+1 age = a . (13.27) numbers are proportional to the catch per unit of effort (CPUE). Actually, equation (13.22) is This gives the probability that an age group of fish equivalent to has a length belonging to length group i. For the weight at age we get y-1 C ay, È ˘ =-qRyaexp Z (13.24) E - Í Â k ˙ y Î kya=- ˚ AWKia, =£Pr() W i weight < W i+1 age = a . (13.28) so that the equation essentially says that the CPUE Keys like the ALK could be applied almost is proportional to stock numbers. The most prob- everywhere in the models, and they would always lematic element in this appealingly simple model represent a more realistic description of the real is the definition and measurement of effort. Total world, but the cost would be a very large number of effort may be composed of effort from many differ- parameters and cumbersome models. A spread- ent vessel types and sizes, all of which can be mea- sheet implementation of a fishery model, on Excel sured in many alternative ways such as sea-days, for example, can use ALKs easily. fishing days, trawling hours and number of hooks. Usually, one needs to keep track of the body Furthermore, due to technical development, one weightofallthecohorts,soanextra index is needed, unit of effort in year y may be different from a unit La,t and wa,t, the length and weight at time t of age of effort in, say, year y -5. For this reason equation groupa. The body sizes at age may be assumed to re- (13.18) is problematic to apply in practice. For main constant from year to year, but they may also further details of this approach see Fournier and vary. In that case one more index is needed, wy,a,t: Archibald (1982). the weight of age group a at time t in year y. 280 Chapter 13

In ICES, size-based models are assumed to be 1 Mata = MM ; the same as age-structured models. The entire the- 1++◊exp()SSL12a ory of size-structured models may be reformulated M ML50% with body length as the basic observation (Chapter S1 = ln()3 ; ML75%%- ML 50 9, this volume, and Sparre and Venema 1992). This SM approach is particularly useful in the tropics where SM =- 2 . 2 ML (13.31) it is more difficult to read age from otoliths. In 50% addition, in all areas, shellfish do not have otoliths. For these and other species groups, ML50% and ML75% is the length at which 50% and length-based models are more useful (Pitcher, 75% of the fish are mature, respectively. Chapter 9, this volume). N SSByt,,,,= Â N yat W at Mat a. (13.32) a=0 13.6.4 Biomass of stock, catch, discards and landings and Where there are large differences between the value of landings sexes, we may need to have two models, one for females and one for males. If the maturity ogive is The von Bertalanffy equation has only three differentiated between sexes, all other equations parameters, whereas each age and size class in an (for example, stock numbers) must also be separate age–length key is a parameter. For this reason, for females and males. careful thought needs to be given before using an A more realistic approach would be to use a age–length key. length–maturity ogive to estimate the proportion The stock biomass at time t in year y may be of fish aged a that were mature. This empirically derived from the one-to-one equation: based estimate could then be used to calculate spawning stock biomass. This method should only aa= max be used if the extra complication yields results BNwyt,,,,= Â yat at, (13.29) a=0 which are a significant improvement on the sim- pler solution shown in equation (13.32). or it may be derived from the age–weight key, The catch may be divided into landings and which illustrates the extra computations needed discards, which implies a model for discards. The when using such a key: logistic curve can also be used here to model the fraction of fish caught, and which are discarded N NW D WWii+ +1 ( a), as a function of body length: Byt= N yat AWK ia , (13.30) ,,,,ÂÂ 2 a=0 i=1 1 Da =-1 DD; 1++exp SSLa where N is the number of age groups and NW is ()12 the number of weight groups. To compute the D DL50% S1 = ln()3 ; spawning stock biomass (SSB), a relationship is DL75%%- DL 50 D needed between either age or length or weight, D S2 S2 =- . (13.33) and maturity. For that purpose we might go for DL50% the easy solution and use a one-to-one relation- ship between length (or age) and the fraction of DL50% and DL75% is the length at which 50% mature fish. We may use the logistic curve to esti- and 75% of the fish caught are discarded, res- mate the proportion of mature fish, which was also pectively. Note that (1 - Da) is the fraction landed. used for trawl gear selection earlier (equation The total quantity discarded by a fleet DFl is 13.16). given by Choosing the Best Model for Assessment 281

N when the relevant data are available, for example, DCwD. Fl= Â Fl, a a a (13.34) through logbook schemes. Some of the problems a=0 faced in stock assessment might be reduced by the To model the effect of a minimum landing size on introduction of models accounting for geographi- discarding we could also use the logistic curve, so cal and seasonal aspects. illustrating that there are many different applica- Geographical aspects may be dealt with in a tions of this curve in fisheries. simple manner by introducing a finite number of The SSB used as a measure of the performance areas, and seasonal aspects by introducing a finite of the system and the level of discards can be added number of seasons. These could be defined in to the list of outputs. Discarding is a political issue weeks, months or quarters of the year. As an ap- as the rejected fish are considered a waste of poten- proximation we then assume that within a time tial catches (Kaiser and Jennings, Chapter 16, this period, all parameters remain constant, fish are as- volume). Therefore discards can be considered as a sumed to be homogeneously distributed within an part of output. This illustrates that there are no area, and nothing moves between the areas during uniqueness rules for nominating certain variables a time period. All movements are assumed to as output, and others as variables. occur at the end of the time period, and take no Fisheries landing statistics are not given by age time to happen. Although an approximation to the group but by commercial size category. The com- real world, any precision can be modelled by se- mercial size categories may be based on body lecting areas and time periods at a scale that can weight, body length or some other criterion, such be supported by the data. For a detailed discussion as the number of tails per pound for shrimps. These of the migration of commercial fish species, see groupings can be transformed into age groups or Metcalfe et al., Chapter 8, Volume 1. size groups by keys similar to the ALK or they may The Migration Coefficient, MC, from area A to be modelled under the assumption of a one-to-one area B is defined as the fraction of the animals in relationship using a growth equation and the area A which moves to area B. In this definition, length–weight relationship. the ‘movements’ include the ‘move’ from area A to The real choice to make is not so much the area A, which covers the case where the animals do mathematical model, for example the logistic not move. The migration coefficient depends on model for gear selection, but rather whether or not the starting area (FAr) and the destination area (TAr). to include a submodel for gear selection. The gear Note that the sum of migration coefficients over selection must be included if the objective of the destination areas is always equal to 1.0, as the model is to predict, for example, the effect that starting area is also considered a destination area: mesh-size regulations might have. In other cases N one may have a choice of whether to include the MC 10., Â aF,,,Ar T Ar q = (13.35) submodel. Ar=1

13.6.5 Models for spatial and where a = age group, q = quarter or any time period. seasonal aspects of fisheries For a theoretical discussion of migration in con- nection with age-based fish stock assessment Stock abundance and consequently catch rates the reader is referred to Quinn et al. (1990). These may vary from area to area. For this reason fishing authors also discuss the estimation of migration effort will also vary between areas, and distribu- parameters. There are also more sophisticated tion of resources and fishing areas will vary be- models, based on differential equations (Olivier tween seasons. Most management agencies do not and Gascuel, 1999). account for seasonal and geographical variations The concept of mortality, as it is traditionally but we will introduce here simple models dealing applied, refers to the entire stock, not to a fraction with areas and seasons. These can be implemented of a stock. Consequently, the number of deaths in a 282 Chapter 13

Recruitment

Area one (Ar = 1) Area one (Ar = 1)

N( y, Ar = 1,q) N1( y, Ar = 1,q) N( y, Ar = 1, q+1) N1( y, Ar = 1, q+1)

a = 0 death process a = 0 a = 0 death process a = 0

a = 1 death process a= 1 a = 1 death process a = 1

a = 2 death process a = 2 a = 2 death process a = 2

Fig. 13.3 The flow of stock

Area two (Ar = 2)Migration (takes zero time) Area two (Ar = 2) numbers (N(year, area, quarter) = N(y,Ar,q) depicted as small N( y, Ar = 2,q) N1( y, Ar = 2,q) N( y, Ar = 2, q+1) N1( y, Ar = 2, q+1) boxes), between model- components from quarter q to a = 0 death process a = 0 a = 0 death process a = 0 quarter q + 1 (during quarters 1, 2 and 3) and between two areas (depicted as large grey boxes) a = 1 death process a = 1 a = 1 death process a = 1 in a hypothetical example with one stock and three age groups a = 2 death process a = 2 a = 2 death process a = 2 (a = 0, 1 and 2). Migrations and recruitments are depicted Transition between quarters as thin arrows and the death (except for transition between quarter 4 and quarter 1) process by fat arrows. For further explanation see text. sub-area of the total area occupied by a stock, To explain the multi-area version of the expo- should not be associated with a mortality, but nential decay model (see equation 13.10), we use with some other concept. Here we use the term the quarter as the time period. The stock at the end area-specific mortality, as the concept required of quarter q of year y in area Ar (before migration) is to describe the death process within a sub-area. modelled by: Area-specific mortality is closely related to mor- tality as normally understood. Let Zstock,y,a,q indi- NN1y a Ar q=- y a Ar qexp Zt y a Ar qD . cate the traditional total mortality of the stock. ,, , ,, ,() ,, , (13.37)

The relationship between Zstock,y,a,q and the area- specific total mortalities, Zy,a,Ar,q, is Note that the indices of N and N1 remain un- changed when considering the death process Ï NA ¸ during a quarter of the year. In the model the Ô Â exp()-ZtNyaArq,, ,D yaArq ,, , Ô Ô Ar=1 Ô transition between quarters occurs between just Zstock,yaq , , =-lnÌ NA ˝ before migration and just after migration. The use Ô N Ô Ô Â y,, a Ar , q Ô Ó Ar=1 ˛ of indices in relation to the transition between (13.36) components is illustrated in Fig. 13.3 for quarters 1, 2 and 3. where Ny,a,Ar,q = stock number at the beginning of The number at the beginning of quarter q in quarter q of year y in area Ar and NA is the number year y in area Ar (just after migration, see Fig. 13.3), of areas. if q = 1,2,3 is given by Choosing the Best Model for Assessment 283

NA fish (Myers, Chapter 6, Volume 1). This is a NMCN1 y,, a TAr , q+1 = Â a , FAr , TAr , q St ,,, y a Ar , q (13.38) central problem of fish population dynamics, since FAr=1 it represents nature’s regulation of population size, whether or not the populations are being if q = 4 (and a < a ) then max exploited. Obviously, there can be no recruits if no adult fish are left to mature, spawn and produce NA eggs, which hatch and grow to become recruits. NMCNy++11,, a TAr , 1= Â a , FAr , TAr , 41 y ,,, a Ar 4 (13.39) FAr=1 Unfortunately, this is about the only thing we know for sure about the stock–recruitment rela- tionship (for a discussion see, for example, Gilbert where: amax is the oldest age group. 1997, Hilborn 1997, Myers 1997 and Myers, Chap- MCa,Ar1,Ar2,q = the migration coefficient for stock St moving from area Ar1 to area Ar2 in quar- ter 6, Volume 1). The problem of recruitment is so important ter q of age group a and N1St,y,a,Ar,q = the number of fish in stock St, at the end of quarter q of year y in that it has a chapter to itself (Myers, Chapter 6, area Ar (before migration). Volume 1). It is difficult to recommend any of the Stock number at the end of quarter q of year y in stock–recruitment models, as very few of them area Ar (before migration) in the oldest age group have been shown to provide a convincing fit to observations. Many fisheries scientists use the amax, which is here modelled as a plus group, that Beverton and Holt model, when a stock and re- is, it contains all the age groups, (amax, amax + 1, amax cruitment model has to be used, but there is no jus- + 2, ...amax +•). If q = 1,2,3 and a = amax then the equation is the same as for the younger tification for this use other than familiarity. Perhaps the best approach is to use ‘no model’, age groups. Only if q = 4 and a = amax is there a dif- ference compared to the younger age groups: meaning that the model is replaced by a stochastic factor multiplied by a constant. Such an approach NA does not pretend to describe a biological mecha- NMCN1 y+11,,, amax TAr= Â a max ,,,,,, FAr TAr 4 y a max Ar 4 nism, which we do not yet understand. Examples FAr=1 NA of stock–recruitment curves and a discussion of MC N1 . (13.40) mechanisms of density-dependent mortality are + Â amax--14,,, FAr TAr y , a max 14 ,, FAr FAr=1 given in detail by Myers (Chapter 6, Volume 1).

Spatial features in general involve mapping and other means of presentation, as well as spatial 13.7 A KEY TO MODELS analysis, which all falls under the category of GIS (Geographical Information Systems). However, In Table 13.1 we present a key to some of the GIS is too big a field to be covered here and the in- main models used in fisheries. This key is far from terested reader is referred to the literature on appli- exhaustive, and we wish to emphasize that while it cation of GIS in fisheries (Castillo et al. 1996; Fox lists major techniques, no model can be applied and Starr 1996; Meaden and Do Chi 1996; Meaden without thinking very carefully about the model’s and Kemp 1997; Starr and Fox 1997; Denis and assumptions and applicability to each particular Robin 1998; Foucher et al. 1998). situation. We refer readers to the relevant chapters of this book as well as Quinn and Deriso (1999) for 13.6.6 Some stock–recruitment detailed explanations of these assumptions as models well as full mathematical derivations. A recent textbook by Jennings et al. (2001) also provides an The stock and recruitment problem is the search overview of methods. As noted earlier, some of the for a relationship between parental stock size and models presented here may meet the objectives of the subsequent recruitment in numbers of young giving advice for fishery management by them- 284 Chapter 13 th the , this chapter s 8, this chapter Shepherd and Pope auxilliary (e.g. stock and catch-at-age models Statistical this chapter + Key to major groups of fisheries models, including primary chapters of this volume where described (John D. Reynolds assisted wi CPUE and effort straight-line regression? by linked CPUE and effort exponential curve? negative by linked CPUE and effort non-linear curve? by linked in model and not data?All error in dataAll error and not in model? model production Fox model Schaefer production years? previous Data one location over from model production and Tomlinson Pella locations?Data many from Aim to examine multiple species caught in same gears? model production Process-error model production Observation-error Schnute and Richards Schnute and Richards Schnute and Richards stat. catch-age models Multi-species yield-per-recruit, model multispecies production Temporal 8 Shepherd and Pope Schnute and Richards Schnute and Richards model Spatial multispecies production Jennings et al. (2001) Aim to examine predator–prey interactions among species?Aim to examine predator–prey models Trophic Pauly and Christensen recruitment)? with fishery? in equilibrium Stock (Be careful!) with fishery? not assumed to be in equilibrium Stock Data on catches and effort only? Data on catches, effort, stock and recruitment? age and size composition, Historical data on catches and age composition?Historical data on catches, age composition cohort analysis VPA, 7 Shepherd and Pope, Data on catches, effort, stock and recruitment? age and size composition, dynamic pool, stat. Yield-per-recruit, catch-age model Aim to forecast? Aim to Data on catches and effort only? model multispecies? Aim to past events?Aim to reconstruct Aim to forecast? MSVPA 7 Shepherd and Pope Table 13.1 Table development of this key). Goals and data available model individuals? Aim to fishers?Aim to model interactions between fishes?Aim to model interactions between species? model a single Aim to past events? Aim to reconstruct individual-based models Game theory, individual-based models Game theory, Huse et al. Huse et al. Model Primary chapter Choosing the Best Model for Assessment 285 selves, or they may be submodels of others in the tions’, principally through predator–prey relation- key. ships. For basic biology of foraging and predation, The first triplet presents a choice between see the trio of chapters in Volume 1 by Mittelbach models of individuals, single species and several (Chapter 11), Juanes et al. (Chapter 12) and Krause species. Most models in fisheries science have et al. (Chapter 13). For ecosystem interactions and dealt with either population dynamic models of impacts see Polunin and Pinnegar (Chapter 14, species or, more recently, the population dynamics Volume 1), Persson (Chapter 15, Volume 1), Jones of many species. Population dynamics treats indi- et al. (Chapter 16, Volume 1) and Kaiser and vidual animals as replicates of each other and Jennings (Chapter 16, this volume), and for incor- models accounting for changes in numbers with poration into fisheries models, see Shepherd and time only recognize reproduction and death as Pope (Chapters 7 and 8, this volume) and Pauly and individual properties. Even then, each individual Christensen (Chapter 10, this volume). has the same chance of giving birth and dying. The Many models aim to reconstruct past stock development of behavioural ecology over the past sizes and fishing mortality, in order to relate these 25 years has shown the importance of recognizing to fisheries effort and hence predict the future. The that two individuals in a population are not the simplest method is virtual population analysis, same and within a population there can be more which is often implemented using a computa- than one strategy for reproduction, feeding and use tional simplification called cohort analysis (Shep- of space (Krebs and Davies 1997). This conceptual herd and Pope, Chapter 7, this volume). Cohort advance, coupled to the power of modern com- analysis may be based either on age classes or on puters, makes it possible to develop models that length classes if age is not known. VPA/cohort are based on the actions of individuals. Three analysis requires initial guesses of fishing mortali- powerful methods have emerged that simulate the ty and natural mortality, and while these esti- way in which organisms have adapted themselves matesbecomelessimportant to the reconstruction to their environment or interact with each other. of stock sizes and fishing mortality in early years, Genetic algorithms mirror many of the processes there is considerable uncertainty in estimates for that occur during natural selection and neural net- the most recent years (Megrey 1989). Statistical works mimic the way in which nervous systems catch-at-age models, discussed in this chapter, are learn about the world. Game theory handles the more sophisticated but also more difficult to im- way in which the reward gained by an individual plement. Single-species VPA has a multispecies when interacting with others depends on the analogue (MSVPA) (Shepherd and Pope, Chapter 7, strategy used by it and its opponent. The first entry this volume), which has a voracious appetite for in Table 13.1 recognizes that these new methods data concerning predatory habits of fishes (Sparre are now being applied to problems of interest to 1991; Magnússon 1995; Christensen 1996). fisheries biologists but at present the approach is The key in Table 13.1 offers two main routes not used at the stock assessment level. Fisheries to forecasting stocks and yields in single-species scientists who still have to confront the reality of assessments. One can collapse the effects of age delivering a stock assessment to management structure and body growth rates, although they are authorities may still be reluctant to use the models still implicitly incorporated by summing over outlined by Huse et al. (Chapter 11, this volume) all age groups, and use production models to assess but they should certainly look to the area as one the behaviour of the entire stock to previous from which a totally new set of methods may fishing effort (Schnute and Richards, Chapter 6, emerge over the next 20 years. this volume). Then one can predict how both yields Multispecies models may account for ‘techni- and stock sizes will be affected by fisheries. In the cal interactions’, due to more than one species past, researchers were willing to make the assump- being caught by the same gear (e.g. Misund et al., tion that such data were derived from stocks that Chapter 2, this volume), or ‘biological interac- were in equilibrium with fisheries. However, this 286 Chapter 13 is a very dangerous assumption, as fishing effort There are multispecies counterparts to the usually increases steadily during the early years single-species forecasting methods (Daan and of a fishery, without giving stocks a chance to catch Sissenwine 1991). Thus, multispecies production up through density-dependent responses (Hilborn models can be used when species are aggregated and Walters 1992). More modern methods, re- in catch statistics. Comparisons are then made viewed by Schnute and Richards (Chapter 6, this between an index of fishing effort and catches in volume) overcome this assumption. Regression- either multiple regions (spatial multispecies based approaches which assume that the error is in production models) or over multiple years from the underlying model rather than in the data a single site (temporal multispecies production (process-error methods) tend to be less precise than models). If numbers and sizes at age are known, methods in which the error is assumed to be in the then multispecies yield-per-recruit and statistical data (observation-error; Polacheck et al. 1993). For catch-at-age models can be used. While multi- certain purposes, the surplus production models species models represent a conceptual advance may perform as well as the sophisticated age-based over single-species models, the difficulty of models discussed below. It is not generally the case gathering data to parameterize them has prevented that complicated models with many parameters their widespread adoption (Evans and Grainger, have higher predictive power than do simple mod- Chapter 5, this volume; FAO 1999). els. This may not seem fair to the hard-working de- A further conceptual advance is provided by signers of age-based or size-based models, but if the trophic models, which attempt to include environ- purpose of the model is to predict CPUE, the sur- mental data and aim to model the whole ecosys- plus production model would work probably as tem, in principle starting with the sun’s radiation, well as the Thompson and Bell (1934) model dis- photosynthesis and primary production. One can cussed earlier. But there is an increasing stream of then model the entire transport of biomass be- requests from management bodies for an analysis tween components of the ecosystem and keep of details. An example would be to assess the effect track of the food web, as explained by Pauly and of technical management measures on catches (see Christensen (Chapter 10, this volume). Software Section 13.6.4). These questions cannot be ad- such as Ecopath (Christensen and Pauly 1992; dressed with a surplus production model. If the key Pauly et al. 2000) makes the bookkeeping fairly issue is the discarding of young fish, the model straightforward. This approach builds on a ground- must contain some splitting of the stock biomass breaking model by Andersen and Ursin (1977), by size and some description of discard practice. We which contained all the major components of the recommend that it is best to test both the simple ecosystem. The multispecies VPA was developed surplus production models and the more compli- as a reduced version of this ecosystem. Thus, at cated size-based models and compare and test the some future stage, all the models may be merged performance of each. Whenever the simple model into a single mega model. performs as well as the complicated model, it should be used. Simple forecasting methods that break stocks 13.8 THE IMPORTANCE down into age classes and account for individual OF TESTING MODELS body growth are reviewed by Shepherd and Pope (Chapter 8, this volume). Yield-per-recruit and Fishery science has a long tradition of modelling dynamic pool models can be easily calculated and the dynamics of exploited stocks. It was probably readily understood, and are in widespread use. one of the first areas of ecology where mathemati- Statistical catch-at-age models (outlined in this cal models were used extensively, and Volterra chapter) are more flexible, but they tend to be more was stimulated by market data from the Adriatic difficult for people other than the modeller to fisheries to produce his famous equations relating evaluate. predator and prey abundances (Kingsland 1985, Choosing the Best Model for Assessment 287 and see Chapter 10, this volume, equation 10.2). model describes the true process. This means that The work of Beverton and Holt (1957) was a stun- the scientist is not clear about the process that is ning creation but the models produced were only bringing about the observed events. This could partly confronted by data. Some of the key models, lead to recommendations which would attempt to such as the logistic model of population growth manipulate the wrong variable. For example, in a and the yield-per-recruit model of Beverton and typical tropical country one may observe hundreds Holt (1957), have not been through a proper cycle of stocks of commercial importance, and the con- of prediction and testing against the data. cept of fish stock assessment becomes cloudy. It is Some fishery biologists, such as Walters (1986), impossible to collect data for, say, 200 stocks in a have proposed ways in which management meas- developing country. At times, fisheries biologists ures could be designed so as to gain information can be tempted to question the usefulness of fish about the predictive capacities of models. So- stock-assessment methodology in general as a tool called adaptive management recognizes that it is for fisheries management in tropical countries often not possible to decide which is the correct (Mahon 1997). model of an exploited population. Walters (1986) When choosing a model that relies on the con- proposed that managers should be prepared to use cept of individual stocks with an age structure, it is the fishery to obtain information that would allow obviously important to consider the number of differentiation between models. After an initial stocks as well as the problems involved in the assessment, a number of models would be pro- definition of the stocks. Definition of a stock can posed, each of which would make different predic- be problematic (Begg et al. 1999), and one can ques- tions about how the fishery would influence the tion the usefulness of methods such as VPA when stock. The area of the fishery should then be split faced with a fisheries system with a large number up and different management regimes imposed in of badly defined stocks. Even in temperate and each. This differential exploitation pattern could polar waters, with relatively few stocks, we face se- be used to test predictions deriving from the differ- vere problems in defining the stocks. ent models. Eventually all but one model would be thrown out as being inadequate. A problem with the approach is that managers are reluctant to 13.9 CONCLUSIONS manipulate fisheries in a way that could lose fish- ers money. This discussion points up the fact that This chapter outlines the basic structure of models advising on which model to use has to be done that have been applied to fisheries and discusses with extreme caution. A model needs to be chosen the ways in which models can be selected and because it accounts for the data available, not applied. A key theme is that each situation will because it is well known and mathematically require the fisheries scientist to have a deep tractable. This means that a fishery scientist needs knowledge of the biology of the species or eco- to be intimately knowledgeable about the fishery system in question and have the ability to tailor he or she is working with, know its biology and models to suite the particular conditions. Our create new elements of a model to deal with the discussion of the objectives of models has been particular circumstances observed (Mangel and general but in most cases the output from a model Clark 1988). is in the form of some type of reference point that In general we do not advocate the use of only can be used to design management measures. The one model, but recommend that different models simplest to grasp might be Maximum Sustainable are tested on the same data set (see also Walters Yield, a concept that has been through several 1986; Hilborn and Mangel 1997). A problem often deaths and reincarnations during its existence encountered is that several models can give the (Larkin 1977; Mace 2001; Punt and Smith 2001). same answer (Schnute and Richards 2001) and Reference points are discussed further in Smith there is insufficient data to determine which et al. (1993), Caddy and Mahon (1995), Quinn and 288 Chapter 13

Deriso (1999) and Gislason (1999). A particular version of the model. The deterministic version is problem with fishery models is that they are not mainly used to understand and check the stochas- often tested properly. Models of population tic version. In general terms, match the model to growth, such as the logistic, which underlies the the needs of the user. surplus production model of Schaefer, are rarely tested to determine if they describe properly the way in which biomass changes over time. It is ob- ACKNOWLEDGEMENTS vious from what is known about the population dynamics of animal populations that the logistic We would like to thank Jon Schnute for reading a equation is not a good model, mainly because it is draft of this chapter and suggesting substantial deterministic and does not account for time lags changes, which greatly helped us with our inherent in the life histories of most animal revisions. species. Yet despite this, logistic models are still used in fisheries. Future work must focus on pro- ducing tested models of population dynamics and REFERENCES fishery scientists must wean themselves away from the so-called traditional models. We need to AD Model Builder (2001) An Introduction to AD MODEL escape from the legacy of the past. BUILDER Version 4 for use in Nonlinear Modeling and In this chapter we also try to provide some guid- Statistics. Otter Research Ltd, Box 2040, Sidney, BC, ance for choosing a model. The first step is to check V8L 3S3, Canada. whether the minimum requirement of the objec- Andersen, K.P. and Ursin, E. (1977) A multispecies exten- tive is met by the available data. Then check if sion to the Beverton and Holt theory of fishing, with there is sufficient funding and manpower as well as accounts of phosphorus circulation and primary production. Meddelser fra Danmarks Fiskeri og time available for the collection of additional data Havundersogelser 7, 319–435. specific to the model. If the objectives are not met Baranov, F.I. (1918) On the question of the biological basis or are unachievable then the objectives should be of fisheries. Nauchn.Issled.Ikhtiol.Inst.Izv., 1, 81–128 changed. The maximum amount of information (in Russian). should be extracted from the data and the model Beddington, J. and May, R.M. (1977) Harvesting popula- used must be appropriate for the type and quality tions in a randomly fluctuating environment. Science of data used. It is always important to make the 197, 463–5. Begg, G.A., Friedland, K.D. and Pearce, J.B. (1999) Stock model as simple as possible with the minimum identification and its role in stock assessment and fish- number of parameters and equations, and one eries management: an overview. Fisheries Research 43 should not invent new equations unless they are (1999), 1–8. really needed. Parameters should be estimated by Bertalanffy, L. von (1934) Untersuchungen über die standard methods, preferably using easy-to-obtain Gesetzlichkeiten des Wachstums. 1. Allgemeine commercial software. When a key parameter can- Grundlagen der Theorie. Roux’Arch.Entwick- lungsmech.Org., 131, 613–53. not be estimated, try to make a plausible guess Beverton, R.J.H. and Holt, S.J. (1957) On the dynamics at it rather than ignoring it. All interacting com- of exploited fish populations. Fish.Invest.Minist. ponents of the system should be included. For Agric.Fish.Food G.B.(2 Sea Fish.), 19, 1–533. example, include in the model spatial and tem- Caddy, J.F. and Mahon, R. (1995) Reference Points for poral factors when these are important for the Fisheries Management. FAO Fisheries Technical output. Make the model multispecies and/or Paper, No. 347. Rome: FAO. multi-fleet whenever these components are Castillo, J., Barbieri, M.A. and Gonzalez, A. (1996) Rela- tionships between sea surface temperature, salinity, important for the output, as they usually are. and pelagic fish distribution off northern Chile. ICES Wherever possible use bioeconomics to model the Journal of Marine Science 53 (2), 139–46. behaviour of fishers, such as their discard practice. Christensen, V. (1996) Virtual population reality. Re- Try to make a deterministic as well as a stochastic views in Fish Biology and Fisheries 6, 243–7. Choosing the Best Model for Assessment 289

Christensen, V. and Pauly, D. (1992) ECOPATH II – a soft- Jennings, S., Kaiser, M. and Reynolds, J.D. (2001) Marine ware for balancing steady-state ecosystem models Fisheries Ecology. Oxford: Blackwell Science. and calculating network characteristics. Ecological Kingsland, S.E. (1985) Modeling Nature: Episodes in the Modelling 61, 169–85. History of Population Ecology. Chicago, IL: Chicago Daan, N. and Sissenwine, M.P. (eds) (1991) Multispecies University Press. models relevant to management of living resources. Krause, A. and Olson, M. (2000) The Basics of S and S- Proceedings of a Symposium held in The Hague, 2–4 Plus, 2nd edn, New York: Springer-Verlag. October 1989. ICES Marine Science Symposia 193. Krebs, J.R. and Davies, N.B (eds) (1997) Behavioural Denis, V. and Robin, J.-P. (1998) Present status of the Ecology: An Evolutionary Approach. 4th edn. Oxford: French Atlantic fishery for cuttlefish (Sepia officinalis) Blackwell Science. Copenhagen: ICES. Larkin, P.A. (1977) An epitaph for the concept of maxi- DiIorio, F.C. (2000) SAS Applications Programming: A mum sustainable yield. Transactions of the American Gentle Introduction. Cary, NC: SAS Institute Inc. Fisheries Society 106, 1–11. Doubleday, W.G. (1976) A least squares approach to Littell, Ramon C., Milliken, Georges A., Stoup, Walther analysing catch at age data. Research Bulletin: Inter- W. and Russell W.D. (1999) SAS Systems for Mixed national Commission for the Northwest Atlantic Models. Cary, NC: SAS Institute Inc. Fisheries 12, 69–81. MacLennan, C.N. (ed.) (1992) Fishing gear selectivity. FAO (1999) Guidelines for the Routine Collection of Fisheries Research (Special issue) 13 (3), 201–352. Capture Fishery Data. Prepared at the FAO/DANIDA Mace, P.M. (2001) A new role for MSY in single- Expert Consultation. Bangkok, Thailand, 18–30 May species and ecosystem approaches to fisheries stock 1998. FAO Fisheries Technical Paper, No. 382. Rome: assessment and management. Fish and Fisheries 2, FAO. 2–32. Foucher, E., Thiam, M. and Barry, M. (1998) A GIS for the Magnússon, K.G. (1995) An overview of multispecies management of fisheries in West Africa: Preliminary VPA – theory and applications. Reviews in Fish Bio- application to the octopus stock in Senegal. South logy and Fisheries 5, 195–212. African Journal of Marine Science 20, 337–46. Mahon, R. (1997) Does fisheries science serve the needs Fournier, D. and Archibald, C.P. (1982) A general theory of managers of small stocks in developing countries? for analyzing catch at age data. Canadian Journal of Canadian Journal of Fisheries and Aquatic Sciences Fisheries and Aquatic Sciences 39, 1195–207. 54, 2207–13. Fox, D.S. and Starr, R. (1996). Comparison of commercial Mangel, M. and Clark, C.W. (1988) Dynamic Modelling fishery and research catch data. Canadian Journal of in Behavioural Ecology. Princeton, NJ: Princeton Fisheries and Aquatic Sciences 53, 2681–94. University Press. Gilbert, D.J. (1997) Towards a new recruitment para- Manly, B.J.F. (1997) Randomization, Bootstrap and digm for fish stocks. Canadian Journal of Fisheries and Monte Carlo Methods in Biology, 2nd edn. Texts in Aquatic Sciences 54, 969–77. Statistical Sciences. London: Chapman & Hall. Gislason, H. (1999) Single and multispecies reference Meaden, G.J. and Do Chi (1996) Geographical Informa- points for Baltic fish stocks. ICES Journal of Marine tion System: Application to Marine Fisheries. FAO Science 56, 571–83. Fisheries Technical Paper, No. 356. Rome: FAO. Hilborn, R. (1997) Comment: Recruitment paradigms for Meaden, G.J. and Kemp, Z. (1997) Monitoring fisheries fish stocks. Canadian Journal of Fisheries and Aquatic effort and catch using a geographical information sys- Sciences 54, 984–5. tem and a global positioning system. In: D.A. Hancock, Hilborn, R. and Mangel, M. (1997) The Ecological Detec- D.C. Smith, A. Grant and J.P. Beumer (eds) Developing tive: Confronting Models with Data. Princeton, NJ: and Sustaining World Fisheries Resources: The State Princeton University Press. of Science and Management. Collingwood (Australia): Hilborn, R. and Walters, C. (1992) Quantitative Fisheries CSIRO, pp. 238–44. Stock Assessment: Choice, Dynamics and Uncer- Megrey, B.A. (1989) Review and comparison of age- tainty. London: Chapman & Hall. structured stock assessment models from theoretical Hovgård, H. and Lassen, H. (2000) Manual on Estimation and applied points of view. American Fisheries Society of Selectivity for Gillnet and Longline Gears in Symposium 6, 8–48. Abundance Surveys. FAO Fisheries Technical Paper. Myers, R.A. (1997) Comment and reanalysis: Paradigms No. 397. Rome: FAO. for recruitment studies. Canadian Journal of Fisheries Hoydal, K., Rørvig, C.J. and Sparre, P. (1982) Estimation and Aquatic Sciences 54, 978–81. of effective mesh sizes and their utilization in assess- Olivier, M. and Gascuel, D. (1999) SHADYS (‘simulateur ment. Dana, 2, 69–95. halieutiques spatiales’), a GIS based numerical model 290 Chapter 13

of fisheries. Example application: the study of a marine sequential fisheries models. Canadian Journal of Fish- protected area. Aquatic Living Resources 12 (2), 77– eries and Aquatic Sciences 51, 1676–88. 88. Schnute, J.T. and Richards, L.J. (2001) Use and abuse of Paloheimo, J.E. (1980) Estimation of mortality rates in fishery models. Canadian Journal of Fisheries and fish populations. Transaction of the American Fish- Aquatic Sciences 58 (1), 10–17. eries Society 1094, 378–86. Smith, S.J., Hunt, J.J. and Rivard, D. (1993) Risk evalua- Pastoors, M.A., Rijnsdorp, A.D. and Van Beek, F.A. (2000) tion and biological reference points for fisheries man- Effects of a partially closed area in the North Sea agement. Canadian Special Publications of Fisheries (‘Plaice box’) on stock development of plaice. ICES and Aquatic Sciences 120. Journal of Marine Science 57, 1014–22. Sokal, R.R. and Rohlf, F.J. (1981) Biometry: The Princi- Pauly, D., Christensen, V. and Walters, C. (2000) Ecopath, ples and Practice of Statistics in Biological Research, Ecosim and Ecospace as tools for evaluating ecosystem 2nd edn. San Francisco, CA: Freeman and Company. impacts of fisheries. ICES Journal of Marine Science 57 Sparre, P. (1991) Introduction to Multispecies Virtual (3), 697–706. Population Analysis. Rapports et proces-verbaux des Polacheck, T., Hilborn, R. and Punt, A.E. (1993) Fitting Reunions. Conseil International pour l’Exploration de surplus production models: comparing methods and la Mer. measuring uncertainty. Canadian Journal of Fisheries Sparre, P. and Venema, S.C. (1992) Introduction to and Aquatic Sciences 50, 2597–607. tropical fish stock assessment. Parts 1 and 2. Manual, Punt, A.E. and Smith, A.D.M. (2001) The gospel of maxi- Revision 1. FAO Fisheries Technical Paper, No. 306/1 mum sustainable yield in fisheries management: and 2. Rev. 1. Rome: FAO. birth, crucifixion and reincarnation. In: J.D. Reynolds, Starr, R.M. and Fox, D.S. (1997) Can fishery catch data G.M. Mace, K.H. Redford and J.G. Robinson (eds) Con- supplement research cruise data? A geographical servation of Exploited Species. Cambridge: Cambridge comparison of research and commercial catch data. In: University Press, pp. 41–66. D.A. Hancock, D.C. Smith, A. Grant and J.P. Beumer Quinn, T.J., II. and Deriso, R.B. (1999) Quantitative Fish (eds) Developing and Sustaining World Fisheries Dynamics. Oxford: Oxford University Press. Resources: The State of Science and Management. Quinn, T.J., II., Deriso, R.B. and Neal, P.R. (1990) Migra- Collingwood (Australia): CSIRO, pp. 190–219. tory catch-age analysis. Canadian Journal of Fisheries Thompson, W.F. and Bell, F.H. (1934) Biological statistics and Aquatic Sciences 47, 2315–27. of the Pacific halibut fishery. 2. Effect of changes in Ricker, W.E. (1954) Stock and recruitment. Journal of the intensity upon total yield and yield per unit of gear. Fisheries Research Board of Canada 11, 559–623. Report of the International Pacific Halibut Commis- Ricker, W.E. (1975) Computation and interpretation of sion 8, 49. biological statistics of fish populations. Bulletin of the Walters, C. (1986) Adaptive Management of Renewable Fisheries Research Board of Canada 191. Resources. New York: Macmillan (reprinted 1997 by Schnute, J.T. (1994) A general framework for developing the Fisheries Centre, University of British Columbia). Part 3 Fisheries in a Wider Context

14 Marine Protected Areas, Fish and Fisheries

NICHOLAS V.C. POLUNIN

14.1 INTRODUCTION and monitoring activities address human activ- ities over extensive areas (e.g. King 1995). Fisheries management and marine protected areas MPAs and fisheries management have never- (MPAs), defined as areas of the sea effectively pro- theless increasingly converged in the last few years tected from exploitative impact, have tradition- (e.g. Bohnsack 1993; Botsford et al. 1997). One rea- ally occupied different but overlapping niches son for this is that mobile fishing gears are now within the field of coastal management. A number recognized to have caused physical damage to of differences are evident between them. MPAs fragile marine habitats in many areas (Kaiser and have been established for a broad range of scien- Jennings, Chapter 16, this volume). These impacts tific, economic, cultural and ethical purposes, are likely to be reversed in areas protected from orientated for the most part to marine nature fishing (e.g. Auster et al. 1996, 1997). It is also conservation (e.g. Jones 1994), but strict evalua- expected that through food webs, fishing may have tion of success is difficult because in most cases, profound indirect effects on marine communities the objectives and criteria of success have been (Kaiser and Jennings, Chapter 16, this volume), poorly defined (e.g. Polunin 1990; Hockey and some of which might also be reversed in MPAs Branch 1997). Fisheries management also has mul- (e.g. Pinnegar et al. 2000). The contention that con- tiple potential objectives (e.g. Hilborn and Walters ventional fisheries management science is the 1992; Hall 1999; Jennings et al. 2001), but these are root cause of the global fisheries crisis has led some typically focused on the resources targeted by fish- to advocate MPAs as a form of precautionary ap- ing, and also consider economic and social out- proach to fisheries management (Roberts 1997a), comes of resource use. MPAs are relatively simple but the practical implications of this proposition area-based regulatory measures, whereas fisheries are considerable. For example, what science is re- management spans a whole suite of measures, in- quired to design such areas, and what will they cluding restrictions on gears and catches, and sea- look like? What will be delivered, and what are the sonal closures. Marine reserves are rather few in uncertainties? If MPAs have to be very much number, whether in developed countries such as greater in size, such increase in scale would not be the United Kingdom (Jones 1999) or tropical devel- inconsistent with holistic approaches to fisheries oping regions such as in the Caribbean (Stanley management, such as ‘fisheries ecosystem man- 1995), and they are almost all very small in area. agement’ in Australia (Done and Reichelt 1998 and Worldwide, the median size of MPAs is approxi- see Pauly and Christensen, Chapter 10, this vol- mately 1600ha (Fig. 14.1). In contrast, the fishing ume). In a general sense, MPAs may be one of many which currently causes most concern is a large- kinds of zones in a comprehensive coastal manage- scale activity, and fishery management measures ment plan (OECD 1993; Bentley et al. 1994). 294 Chapter 14

300 is necessary to look at the evidence for such effects. This may allow patterns to be discerned in the 250 information.

200 14.2.1 Abundance and body size of target species 150 Drawing on data from underwater visual surveys

Number of MPAs Number 100 and on fishery catch-per-unit-effort (CPUE) data, and including the case of the North Sea where 50 entire grounds were closed during the First and Second World Wars, evidence for greater abun- 0 dances in MPAs than in unprotected areas is now 10 1000 100000 >1000000 available for a range of resource species in more Area (ha) than 40 locations around the world, but there are significant spatial patterns in this evidence (Table Fig. 14.1 Size distribution of 991 MPAs around the 14.1). For example, evidence for fishes derives world for which area is known. (Source: after Kelleher almost exclusively from underwater visual census et al. 1995, p. 14.) work on site-attached species on reefs in the Caribbean, western Mediterranean, Kenya, South Africa, the Philippines, New Zealand and New One of the main objectives of this chapter is to Caledonia. Jones et al. (Chapter 16, Volume 1) have review the known impacts of MPAs with respect to noted differences in census techniques between fish conservation and fisheries management objec- temperate and tropical reef areas, and noted that tives, and help to characterize where, and if so these hamper comparisons of community ecology how, MPAs may be expected to contribute to the between these habitats. An exception is the evi- achievement of such objectives. I will also relate dence from the North Sea, where information MPAs to broader issues of environmental manage- came from catch and market statistics and the ment, including the uses of coastal waters by area protected spanned complete stocks and the tourists, participatory conservation and the main- duration of both world wars (Borley et al. 1923; tenance of biodiversity. Margetts and Holt 1948). Otherwise, the lack of information on the effects of MPAs or fish abun- dance from South America, West Africa, South 14.2 WHAT HAPPENS TO Asia and the Far East is a prominent feature; one TARGET SPECIES IN MPAs? of the North America East Coast studies is a fish- ing-based investigation of reef fishes in Florida, Fishing is an additional source of mortality, and re- while the other two are equivocal and are complex duces survivorship of target organisms. Thus the assessments of large-scale closures (Table 14.1). principal direct effects of exploitation are reduc- In a meta-analysis of 12 MPAs, selected on the tions in abundance, age and size. Accordingly, basis of full reporting of relevant data, Mosquera increases in mean numerical abundance, size, et al. (2000) showed that an overall positive effect age and biomass, of target species depleted by fish- of protection is discernible in the abundance of ing are expected particularly in MPAs. These fishery target species but not in that of non- effects are not assured. They may be mitigated target species. Where many species have been in- by a number of factors and processes, including vestigated, increase in abundance of target species mobility and spatial variability of recruitment at in MPAs has not been detected in all cases (Table large scales relative to the size of MPAs, so that it 14.1). Species other than fishes which have been Marine Protected Areas 295 . .33 .– .22 .42 .–5 .– 1 .30 .6.7 no information) in abundance of various marine fishery = 21. Munro 1999. 22. et al. 1997. Wantiez 23. 1998. Wallace 24. García-Rubies and Zabala 1990. 25. Polunin and Roberts 1993. 26. Algae: Castilla & Bustamante 1989. 27. Stoner and Ray 1996. 28. Francour 1991. 29. Pipitone et al. 2000. 30. Roberts and Polunin 1992. 31. Borley et al. 1923, Margetts and Holt 1948. 32. Rakitin and Kramer 1996, Chapman 1999. 33. Polunin 1999. 34. 1999. Polunin and Williams 35. Funicelli et al. 1988. 36. Hockey and Bosman 1986. 37. Smith and Berkes 1991. 38. Fogarty and Murawski 1998. 39. Armstrong et al. 1993. 40. Edgar and Barrett 1999. .29 – 2 . – 16.36* – – – 20.40* 14* .28 .24 increase in some species, not others, – = .23. 17.18*.38 – 29 9 . – 35. 31. 4 .34 .32 no increase, _ = .25 1 Changes (increase, or * America America America America Europe Europe Africa Africa Asia Asia Asia Pacific W. South W. E. South Caribbean North W. E. North Western Southern West East South Southeast East Australasia South 1. Russ 1985, Alcala 1988, White and 1990, 1996. 2. McClanahan and Shafir 1990, Samoilys 1988. 3. Clark et al. 1989. 4. Bell 1983. 5. McCormick and Choat 1987, Cole et al. 1990, Babcock 1999. 6. Buxton and Smale 1989. 7. Bennett and Attwood 1991. 8. Davis 1977, and Dodrill 1980, 1989. 9. Cole et al. 1990. Molluscs 15. – 13.27. 1 Decapods 15* – 8. 39* 11. – – – – – – 12. 9.10.40 te2.–––––––––––– UrchinsOther26.–– 15. – 37. – – – – – – – – – – – Fish – – 3.2 10. Bowen and Hancock 1985. 11. Leary 1985, Klima et al. 1986, Roberts 1986. 12. and Kuwahara 1990. Yamasaki 13. and Laughlin 1984. Weil 14. Heslinga et al. 1984. 15. Castilla and Durán 1985, Moreno et al. 1984, Fernandez 1998. 16. Lasiak and Dye 1989. 17. Rice et al. 1989. 18. McCay 1988. 19. 1989, 1992. Tegner 20. Shepherd 1990. Table 14.1 Table target species in areas closed to fishing, or otherwise very lightly fished, in different parts of the world. 296 Chapter 14 found to be more abundant in MPAs than unpro- egg production to be raised. In tropical snappers tected areas include muricid gastropods, abalone, (Lutjanidae), spawning may be more frequent and limpets, lobsters, sea urchins and kelp, where these occur over a longer period in larger than in smaller are exploited; these are from reef habitats. The animals (Grimes 1987), but more data are needed overwhelming evidence is that greater abundances to underpin the supposition that greater fecundi- develop in MPAs when site-attached species in the ties give rise to a greater total egg production in area have been substantially depleted by fishing. MPAs than in unprotected areas. It is reasonable to Evidence for greater average sizes of target expect that in site-attached species which do not species in MPAs than in unprotected areas comes become more vulnerable to fishing through move- almost entirely from the study of site-attached reef ment out of protected areas, depleted local popula- fishes investigated by underwater visual census; tions in MPAs will support substantial increases in many of the studies are the same as those which egg production. have demonstrated abundance effects (Table 14.2). In both reef fish population ecology and fisheries For reef fishes, the evidence is from work in population dynamics, the relationship between the Caribbean, Florida, the Mediterranean, Kenya, any increased egg output and actual recruit- South Africa, the Philippines, New Zealand and ment is uncertain (e.g. Sale 1991; Myers and Australia. Several target invertebrate species have Barrowman 1996; Myers, Chapter 6, Volume 1; also been found to be larger in size within MPAs Jones et al., Chapter 16, Volume 1). The scientific (Table 14.2), and these are again associated with basis for such uncertainty varies between the two reef habitat. fields; to workers on reef fishes, recruitment refers to the supply of settlers from the plankton (though 14.2.2 Fecundity, recruitment and see Hixon 1996) and is measured at very small spa- movement of target species tial scales, while to fisheries scientists it is the abundance of fish of a size large enough to be fished Increase in fecundity with body size is expected to and measured from large-scale fisheries data (e.g. occur in fish and other fishery organisms (Duarte Hilborn and Walters 1992; Myers, Chapter 6, and Alcaraz 1989; Elgar 1990). Given the variabil- Volume 1). Much of the evidence has supported ity of fecundity data and the lesser increase in fe- the origin of recruitment variability lying in the cundity with size in some species (Wootton 1990; egg and larval stages, which in fishes are rarely of Sadovy 1996), population fecundity will not in- less than two weeks’ duration (e.g. Brothers et al. crease appreciably in all species within MPAs, but 1983; Houde 1987; Thresher and Brothers 1989). It greater total fecundities of site-attached fishes has been inferred that in reef fishes, and most prob- should occur in MPAs than in unprotected areas. ably all marine fishes, recruits travel distances There will, of course, be exceptions. from spawning sites that are commonly 25–50 The rate at which population fecundity can be times greater than the small spatial scales of most expected to increase will depend upon the rate of MPAs (Fig. 14.1). These distances may be modulat- growth and mortality and the relationship be- ed by current patterns that will vary greatly over tween fecundity and body size (Sadovy 1996). Be- time. For particular localities, species and seasons cause greatly increased survivorship within MPAs there may be a high likelihood of small-scale local is likely, species so protected should attain maxi- retention of larvae, settlers and recruits (Jones et mal population fecundities within five years or so. al. 1999; Swearer et al. 1999). The extent to which Slower-growing species may take 10–20 years to do increased egg output rates from MPAs will lead to this (Polunin 1997). greater abundances of late juvenile and adult reef If maximal population fecundities are to pro- fishes within particular MPAs remains an open duce a greater egg output rate by the stocks of question. A major reason for this is that the dy- females within MPAs then spawning must be namics of larvae and new settlers are little known. frequent enough in the larger individuals for the In marine animals that do not possess larvae, or Marine Protected Areas 297 . 9 1 . 10*. . – 1. – 3.16.18.23*.32 no information) of various marine fishery target = .5.17.25 18. Beinssen 1989a. 19. et al. 1997. Wantiez 20. 1998. Wallace 21. Polunin and Roberts 1993. 22. García-Rubies and Zabala 1990. 23. Ferreira and Russ 1995. 24. Algae: Castilla and Bustamante 1989. 25. Roberts and Polunin 1992. 26. Borley et al. 1923, Margetts and Holt 1948. 27. Rakitin and Kramer 1996. 28. 1999. Polunin and Williams 29. Funicelli et al. 1988. 30. Hockey and Bosman 1986. 31. Chapman and Kramer 1999. 32. Fogarty and Murawski 1998. 33. Edgar and Barrett 1999. increase in some cases, not others, – = . 13.14*.32. – – – 12.30. – – – 15.32 no change, _ = . – 29. 26. 2.22. – 4 1 .3 . – – – – ––– – 8.32.– .28 .27 1 Changes in average size (increase, * America America America America Europe Europe Africa Africa Asia Asia Asia Pacific W. South W. E. South Caribbean North W. E. North Western Southern West East South Southeast East Australasia South 1. Russ 1985, Alcala 1988, and 1990. 2. Bell 1983, Francour 1991. 3. McCormick and Choat 1987, Cole et al. 1990, Babcock 1999. 4. Buxton and Smale 1989. 5. Bennett and Attwood 1991. 6. Davis 1977. 7. Klima et al. 1986, Roberts 1986. 8. and Kuwahara 1990. Yamasaki 9. and Laughlin 1984. Weil Decapods 11* – 6.7 te2.–––––––––––– UrchinsOther24.–– 11. – – – – – – – – – – – – – Molluscs 11. – 9. 20 Fish – – 2 Table 14.2 Table species in areas closed to fishing different parts of the world. 10. Heslinga et al. 1984. 11. Castilla and Durán 1985, Moreno et al. 1984, Fernandez 1998. 12. Lasiak and Dye 1989. 13. Rice et al. 1989. 14. McCay 1988. 15. Shepherd 1990. 16. 1986. and Ayling Ayling 17. McClanahan and Muthiga 1988. 298 Chapter 14 have very short egg and larval stages, build-up of be expected to increase as body size of depleted spawning stock biomass in an MPA may well lead target species increases in MPAs. For territorial to greater recruitment in the MPA, but in fisheries species in MPAs of given sizes, it will be possible such cases are rare (e.g. Carr and Reed 1993; for individuals to move across boundaries into Allison et al. 1998). In the Discovery Bay Fishery similar habitat where there is fishing. Movement Reserve, Jamaica, where target species have been may, however, be curtailed by habitat discontinu- severely fished down, nearly all the larval settle- ities; in the case of reef fishes, reef edges and open ment was by species with extended pelagic phases sand constitute such barriers to movement. Such (Munro and Watson 1999). In areas lacking such discontinuities should therefore help promote stock depletion, local recruitment which is less biomass accumulation in MPAs (Kramer and reliant on the vagaries of larval and early post- Chapman 1999). settlement life is a possibility. It is possible to say, that in areas with extensive and severe depletion, the scope for local recruitment will have been sub- 14.3 WHAT ARE stantially lost. It must be said that, larval supply is THE POTENTIAL BENEFITS not the only process governing the abundance of animals of exploitable size; for example, abun- TO FISHERS? dance of juvenile gruntfish (Haemulidae) has been found to be lower within the Barbados Marine Re- Even in the case of sedentary species in reef habi- serve than outside it and this is attributable to the tats, fishers may be disadvantaged by MPA cre- greater density of piscivorous fishes as a result of ation because they lose part of their ground and protection (Tupper and Juanes 1999). thus access to part of the stock, and total catches The spatial scale of the movement of species may therefore decline when MPAs are put in place that are targeted by a fishery will clearly influence unless total fishing effort is reduced (Beverton the likelihood that abundance will increase in and Holt 1957, p. 393; Horwood et al. 1998). A MPAs (Kramer and Chapman 1999). In the reef- short-term decline may ultimately be made good associated species which thus far have best exhib- through recruitment and/or movement effects, ited abundance and size increases in MPAs, home- the latter being referred to as ‘spillover’ (Rowley ranging and territorial behaviour appear to be very 1994). It is therefore appropriate to assess the evi- common. The spatial scale of routine movement is dence for recruitment and spillover before review- indicated by home range and territory size, and ing the evidence for impacts of MPAs on actual this is expected to increase exponentially with in- catches. crease in body length. An indication of the positive relationship between mean home range area and 14.3.1 Spillover and mean fork length for 15 reef fish species is given by recruitment effects Kramer and Chapman (1999). There seem to be no data on how home range area might vary with size In contrast to the number of studies on abundance within important target species, such as the snap- and size effects, few studies have addressed the pers (Lutjanidae) and groupers (Serranidae) of tropi- movement of target species across the boundaries cal reefs. It is reasonable to suppose on energetic of MPAs (Table 14.3). Spatial differences in size and grounds that as body size increases in MPAs, so abundance have been used to infer movement, but fish will tend to range over greater distances. tagging and tracking can alone indicate, the extent Among species in temperate waters, the plaice and direction of movements. Holland et al. (1993, Pleuronectes platessa exhibits a linear relation- 1996) used both to infer the extent of movement ship between migration distance and body length by the trevally (Caranx melampygus) and goat- in Dutch waters (Rijnsdorp and Pastoors 1995). fish (Mulloides flavolineatus) in and out of the This means that the probability of movement can MPA around Coconut Island, Hawaii. Funicelli Marine Protected Areas 299 b movement in some species not = data from modelling only) in different parts of the = b 8–. – 6 –8*–7.– a 15. Bryant et al. 1989, Rutherford et al. 1989. 16. Buxton and Allen 1989. 17. Beinssen 1989b, 1990. 18. Hunte and S. Scott, unpublished data. W. M. Corless, B. Hatcher, 19. Attwood and Bennet 1994. 20. Davies 1996. 21. ICES 1990. 22. Bailey 1991. 23. ICES 1999, Pastoors et al. 2000. 24. Beverton and Holt 1957, Horwood et al. 1998. 25. Munro 1999. 26. Kelly 1999. 27. 1996. Watson 28. Daan 1993. 29. Quinn et al. 1993. b .28 increase under some conditions, not others) of various marine fishery target b = .24 4.21.22.23* 5 no movement across MPA boundary recorded, _ no movement across MPA = b 0 – – – ––– – – – – no change, _ = – 1.15.–1 – – – 16*.19.27* – – – 17.20. 2. b seasonal only) either in practice or inferred from simulation ( = a Movement of target species in and out of MPAs (* Movement of target species in and out MPAs America America America America Europe Europe Africa Africa Asia Asia Asia Pacific W. South W. E. South Caribbean North W. E. North Western Southern West East South Southeast East Australasia South S. Mangi, unpublished data 1999. 1. Funicelli et al. 1988. 2. Holland et al. 1993, 1996. 3. Klima et al. 1986, Gitschlag 1986. 4. Borley et al. 1923, Margetts and Holt. 1948. 5. Garcia and Demetropoulos 1986. 6. DeMartini 1993. 7. Alcala and Russ 1990. 8. McClanahan and McClanahan and Kaunda-Arara 1996, T.R. 9. Sladek Nowlis and Roberts 1999. DecapodsYield –Fish – – 3.13. – – 9 – – – – – – – 11. 26. – 10. Polacheck 1990, Guénette and Pitcher 1999. 11. and Kuwahara 1990. Yamasaki 12. Klima et al. 1986. 13. Davis 1977, and Dodrill 1980, 1989. 14. Chapman and Kramer 2000. Spillover Fish – – 14.18*.25 species in areas closed to fishing ( others) and changes in overall fishery catch (increase, * Table 14.3 Table world. DecapodsUrchins – – – – – – – 29 12* – – – – – – – 26* – 300 Chapter 14 et al. (1988) tagged mullet (Mugil cephalus), drums is stock dependent (Plan Development Team 1990; (Sciaenops spp.) and spotted seatrout (Cynoscion Ault et al. 1998) and will therefore tend to increase nebulosus), while Rutherford et al. (1989) tagged as biomass accumulates within MPAs, but clearly grey snapper (Lutjanus griseus), to infer move- this will depend upon the species being relatively ments within and out of MPAs on the east coast sedentary (e.g. DeMartini 1993). The actual range of Florida. Other studies have demonstrated over which such recruitment can be expected to movements of decapods, including crabs in Japan occur is a matter for debate. In a small number of (Yamasaki and Kuwahara 1990), and lobsters in marine fishes (e.g. the scorpaenid Sebastes and Florida and New Zealand (Davis 1977; Davis eelpout Zoarces) and invertebrates with direct de- and Dodrill 1980, 1989; Kelly 1999), and shrimp velopment, or very limited planktonic lives, the (Klima et al. 1986; Gitschlag 1986) in Florida. The range should be small, but in most cases the fish re- evidence from such studies is that movement cruitment supported by MPAs maybe distributed does occur across MPA boundaries, but that the over tens to hundreds of kilometres. The shape and distance over which it occurs routinely in reef- orientation of the dispersal envelope will depend associated species tends to be small; most informa- on species, season and other factors (Roberts tion indicates spatial scales of less than one kilo- 1997b). The larval longevity of most fishes is prob- metre. In other species, such as mullet (Funicelli et ably best viewed as a measure of the range of dis- al. 1988), or most commercial species in the North tances over which dispersal can occur, because of Sea, it is evident that the spatial scale of movement other factors, such as current speed and direction can be much greater, although it should be noted which will affect dispersal in practice (e.g. Carr and that in reef species the scale of movement may be Reed 1993). greater than indicated by the short-term studies above. This movement can occur in a variety of 14.3.2 Impacts on fisheries yield contexts, including home-ranging and foraging behaviour (Holland et al. 1996), spawning migration The evidence for greater yield in fisheries adjacent (Holland et al. 1993), and changes in habitat as to MPAs is based on few studies, in contrast to the animals get older (Parrish 1987). The distance range evidence for greater abundance and body size of of movement in reef-associated species is similar target species in MPAs (Table 14.3). The empirical to that of their home ranges, but such data probably work has been on reef fisheries in Kenya and the do not include seasonal movements. These move- Philippines (Alcala and Russ 1990; McClanahan ments include those for spawning which occur over and Kaunda-Arara 1996), and demersal shelf spe- greater distances and may be important as the sus- cies such as plaice in the North Sea (Borley et al. ceptibility of species to fishing is likely to be greater 1923; Margetts and Holt 1948), and shrimp in in some species during spawning (Fulton et al. 1999). Florida (Klima et al. 1986). McClanahan and The build-up in spawning biomass which oc- Kaunda-Arara (1996) showed that although catch curs in site-attached species in MPAs may lead to per unit effort increased within 2km of the bound- greater egg output, but the consequences of this for ary of the Mombasa Marine Park, total catch recruitment are uncertain and there appear to be declined because the MPA consituted a large part no data indicating greater availability of fish to any of the ground and fishing effort per unit area in- fishery as a result of the greater supply of larvae. creased; seven years after closure, the catch was Assuming a metapopulation structure such as still reduced (T.R. McClanahan and S. Mangi, un- may occur in some arrays of patch reefs over dis- published data). The yield of pink shrimp (Penaeus tances of tens of kilometres, it is likely that MPAs duorarum) following establishment of a fishery will enhance yield especially when the fishing MPA in the Tortugas of Florida was greatly affected mortality is high (Man et al. 1995). In some cases, by recruitment, but compliance by fishermen with such as in the southeastern USA, the spawning the closure was also poor (Klima et al. 1986). One of stock biomass may be so depleted that recruitment the most striking examples of the effects of closure Marine Protected Areas 301 are those of the ‘great fishing experiments’ based (1999) included both Schaefer and Beverton–Holt on fisheries yield before and after both world wars stock–recruitment relationships in a model as well in the North Sea (Smith 1994 and Smith, Chapter as transfer across a hypothetical MPA boundary, 4, this volume). Although there are problems with they showed that at the fishing mortality corre- interpretation of the effort data, the evidence for sponding to maximum sustainable yield (MSY), these demersal fisheries is that with large-scale the yield is likely to be less when part of the ground closure of entire grounds tens to hundreds of kilo- is protected from fishing. However, when fishing metres across, substantial increases in catch rate mortality was double that producing the MSY, and total catch occur over time-spans of much less the yield was maintained in the presence of the than 10 years (Borley et al. 1923; Margetts and Holt MPA and exceeded that without a protected area 1948). Other more recent examples of large-scale (Guénette and Pitcher 1999). The suggestion is closures in the North Atlantic include those of that the recruitment effects of MPAs will be more herring (Clupea harengus) and capelin (Mallotus important for enhancement of yield than the villosus) (ICES 1990; Bailey 1991), but the North spillover effects, but as yet there are no data about Sea plaice (Pleuronectes platessa) box had failed to recruitment dynamics at the scale typical of MPAs. produce an increase in yield or stock biomass four Intense fishing should often increase year-class years after permanent closure following earlier variability by making stock biomass more prone to seasonal closure (ICES 1999). Another example of a annual recruitment (but see, e.g. Rijnsdorp et al. large-scale temperate demersal fishery closure is 1992). As a result MPAs may under certain circum- that of the extension of the seasonal ban on trawl- stances reduce this variability by building spawn- ing off the coast of Cyprus, where the MPA sub- ing stock biomass. Only modelling results are stantially increased yield within two years of the currently able to provide specific predictions of the extension because the fishing mortality on fish in extent to which this will happen. Thus in a simu- the first few months post-recruitment had prob- lated cod (Gadus morhua) stock, the number of ably been very high (Garcia and Demetropoulos years with poor recruitment is likely to be reduced 1986). when some 10% or more of the ground is closed to In the absence of extensive data both on spill- fishing (Fig. 14.2). Using a simple logistic model, over and recruitment through increased egg and Lauck et al. (1998) have shown that the likelihood larval output, modelling studies have provided of maintaining a stock above, say, 60% of carrying predictions of yield under different MPA and fish- capacity for at least 20 years, is likely to fall as the ing conditions, taking account of varying biologi- temporal variability in the catch increases. The cal features of the animals involved (Table 14.3). proportion of a ground that must be protected to With increase in MPA size, the spawning stock bio- maintain the stock in this way is likely to be more mass per recruit can be expected to increase, but than 50%. the yield per recruit outside the MPA will not in- crease, unless there is a decline in fishing effort; in 14.3.3 Under what conditions will fact, if the fishing effort remains the same, the yield fish and fisheries benefits be detected? will decline, because the effort per unit of fished habitat will have increased. MPAs can scarcely The science of MPAs will no doubt be long in the be expected greatly to increase yield (DeMartini development. Except in broad outline, extrapola- 1993), except where they are large and fishing tion of data from one site to another is unwise, but mortality is very high (e.g. Russ et al. 1992). The it is reasonable to ask what general factors may simplest yield-per-recruit models do not take ac- affect the likelihood of detecting fish and fisheries count of recruitment variations in relation to stock effects. (Shepherd and Pope, Chapter 8, this volume). This Proper demonstration of fish abundance was true of the models used by Russ et al. (1992) changes or fishery effects of MPAs requires com- and DeMartini (1993). When Guénette and Pitcher parison of the state before and after introduction of 302 Chapter 14

(a) Beverton–Holt Ricker 60 60

50 50

40 40

30 30

20 20

10 10 Years of poor recruitment

0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.1 0.2 0.3 0.4 0.5 0.6 0.7

(b) 600 300

500 250

400 200

300 150

200 100 Spawner biomass 100 50

0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.1 0.2 0.3 0.4 0.5 0.6 0.7

(c) 200 120 Reserve size No reserve 100 150 0.1 80 0.3 0.5 100 60

Yield 0.7 40 50 20

0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Exploitation rate Exploitation rate

Fig. 14.2 Changes in (a) the number of years with poor recruitment, (b) spawner biomass, and (c) yield, of an Atlantic Cod (Gadus morhua) fishery as a function of exploitation rate and the proportion of the fishing ground (‘reserve size’) set aside as an MPA. Exploitation rate u is given by the equation u = 1 - e-F, where F = the fishing mortality. (Source: from Guénette and Pitcher 1999, p. 299.) the management regime and with unmanaged con- taining to yield itself fewer still. Warmer waters trols for comparison. This has seldom been possi- provide the opportunity of direct visual underwa- ble. In both tropical and temperate waters, the data ter surveys (e.g. Polunin and Roberts 1993; Russ on changes in abundance are few, and those per- and Alcala 1996 a point also mode by Jones et al., Marine Protected Areas 303

Chapter 16, Volume 1), but, at least in clear waters Appeldoorn 1996, Chapter 2.6; Schnute and elsewhere, video is being recognized as a means of Richards, Chapter 6, this volume). Yield- discriminating spatial differences in abundance per-recruit models rely on catch, mortality and age (Willis et al. 2000) and size (Harvey and Shortis data (Sparre and Hart, Chapter 13, this volume; 1996) data. In many cases, including many fishery Shepherd and Pope, Chapter 8, this volume). species, abundance and size estimation has to rely However, confident estimation of the underlying on sampling techniques such as fishing gears, growth and mortality data is difficult in many including traps and nets (Jones et al., Chapter 16, areas (Appeldoorn 1996). Such data tend to be even Volume 1). Catch-per-unit-of effort data offer a more highly aggregated biologically and spatially relative measure of abundance, but the precision than the best results from underwater visual work, of this will typically be low (Appeldoorn 1996; and there is therefore a difference in spatial size of Schnute and Richards, Chapter 6, this volume; the sampling unit involved between using yields Sparre and Hart, Chapter 13, this volume). and using visual methods. Given spatial and measurement sources of vari- ability, statistical power for detecting most differ- 14.3.4 When and where can fisheries ences when they exist is small, where statistical benefits be expected to occur? power is the probability of correctly rejecting a null hypothesis of no difference between MPAs Many factors will affect the likelihood of fisheries and unprotected areas. Put another way, only benefits occurring, but given the number of when large differences exist will they be detected unknowns, it is worthwhile focusing on a few (Fairweather 1991). It follows that in many cases, which might act as useful predictors. These factors lesser but significant differences will not be detect- include biological characteristics of the organ- ed. High spatial variability, small sample areas and isms, features of MPAs and their environments, numbers, differences in life history and behaviour, and modes of exploitation. and high measurement error, perhaps caused by One factor affecting MPA success in terms of differences in procedure or acuity between divers, fisheries benefits is the mobility of the species con- help to explain apparent differences among species cerned. As mobility of species is a key determinant in response to the creation of MPAs (e.g. Polunin of the build-up of biomass within MPAs, it is and Roberts 1993). Yet, as noted by the meta- clear that it will affect the likelihood not only of analysis of Mosquera et al. (2000), in MPA assess- spillover effects, but also of the recruitment func- ments to date, even basic statistics such as vari- tions of MPAs. This can be simply expressed by ances in means are sometimes not reported. There comparing species that are highly mobile with has also been very little attention to sources of those that are not (Table 14.4). Where species are measurement error, and full details of sample de- highly mobile, the probability of movement out of sign are rarely given, yet the latter is crucial to MPAs will be high and biomass will not build up defining spatial and temporal variability as a basis within them; this means not only that recruitment for assessing management effects. Differences in effects, will be small, but also that spillover will be abundance between areas subject to different lev- small, on the grounds that it is a function, of els of management have been much more readily biomass and the probability of movement (Table detected when species abundance data have been 14.4). Conversely, for sedentary species, the aggregated into trophic or taxonomic groups (e.g. recruitment function of MPAs will be maximal, Russ and Alcala 1996; Polunin and Jennings 1998). but the spillover effects will be small; on this basis, Assessments of yield from fished areas are pos- the greatest spillover is likely to be brought about sible through several means. Surplus production by species of ‘intermediate’ mobility. This is sup- models are simple, but preferably require long ported by modelling, where the greatest yield per time series including good contrasts between recruit occurred in a fish of intermediate mobility, catches and effort (Hilborn and Walters 1992; although only the most mobile species exhibited 304 Chapter 14

Table 14.4 Consequences of species mobility for recruitment through increased egg output and spillover functions of MPAs in intensively fished areas.

High mobility Moderate mobility Low mobility

Biomass accumulation in MPA Low Medium High Recruitment function Minimal Intermediate Maximal Spillover function Intermediate Maximal Intermediate

increase in yield with increasing MPA size, pro- mackerels (Scombridae) was greater with an MPA vided the fishing mortality was sufficiently high much less than 1km2 in area (c. 25% of the ground) (DeMartini 1993). In the case of Sumilon Island in than without (Alcala and Russ 1990). In temperate the Philippines, increases in yield were supported waters, measurements of yield effects come princi- especially by fusiliers (Caesionidae), but also by pally from MPAs spanning 10000–>500000km2 jacks (Carangidae) and mackerels (Scombridae) (Table 14.3), but results at the lower end of this (Alcala and Russ 1990), which are relatively spatial scale are equivocal, as in the case of mobile species for reefs (e.g. Holland et al. 1996). the North Sea ‘plaice box’, and Cyprus and Other biological characteristics may also help Georges Bank (2500–6000km2) closures (Piet and predict fishing impacts and the likelihood for re- Rijnsdorp 1998), although the seasonal closure in covery from exploitation in MPAs. For the North Cyprus had a dramatic effect on yield (Garcia Sea, age at maturity, maximum size and potential and Demetropoulos 1986). While these data con- rate of population increase are useful predictors of trast the effects of MPAs in reef and temperate dem- fishery depletion (Jennings et al. 1997), while in ersal fisheries, the indication is that absolute size is reef fisheries, maximum size can help predict not a good predictor of fishery effects. Modelling decline in target-fish abundance (Jennings et al. studies help to emphasize that it is the proportion 1999). Where a stock is depleted through direct of whole fishing grounds which is important (e.g. effects of fishing alone, body size may help to pre- Guénette and Pitcher 1999), rather than absolute dict recovery in MPAs. Body size is clearly related size, and, further, that fishing mortality relative to to a number of other characteristics, including po- natural mortality is going to be important. tential longevity, natural mortality, growth rate The time-scales over which MPA benefits can and recruitment, and thus does not of itself provide be expected have been examined in a few empirical an explanation for variations in recovery rate and modelling studies. For some large vulnerable among organisms (Russ and Alcala 1998b). reef predators in the Philippines, spillover effects On tropical reefs, very few studies have exam- may take about ten years to become evident (Russ ined changes over time, and it appears that only two and Alcala 1996). The accumulation of biomass to date have determined effects on yield, and pro- within MPAs of site-attached species vulnerable to duced different results, from intensive fisheries fishing can occur within 10 years (Fig. 14.3) (see (Table 14.3). In Kenya, a fishery with an MPA also Borley et al. 1923; Margetts and Holt 1948; 10km2 in area, which is about 60% of the ground, Polunin and Roberts 1993), if the MPA is large had produced no increase after seven years in a enough to accommodate the mobility and vulnera- total catch contributed to mostly by rabbitfish bility of the species concerned. In spite of the fact (Siganidae), parrotfishes (Scaridae), octopus and that the scientific basis for it is better understood squid (McClanahan and Kaunda-Arara 1996; than for other strategies such as spillover and lar- T.R. McClanahan and S. Mangi, unpublished data val recruitment through spawning biomass accu- 1999). In the Philippines, a reef catch predominant- mulation, pulse fishing of MPAs is considered ly of fusiliers (Caesionidae), jacks (Carangidae) and by some not to be a feasible option. This is on the Marine Protected Areas 305

(a) 20 aesthetic. These criteria have served implicit purposes of attracting tourism, educating the pub- Y = 2.58e0.18X r 2 = 0.89 16 lic or signalling the presence of sensitive commu- nities in particular areas (see below). However, the science of reserve design and location in relation to 12 issues such as larval supply to depleted fisheries has scarcely progressed beyond an elementary 8 Sumilon Reserve modelling stage (e.g. Man et al. 1995; Guénette and Pitcher 1999; Sladek Nowlis and Roberts 1999; )

2 4 Planes et al. 2000; Sánchez Lizaso et al. 2000). Some general guidelines on the design of MPAs can 0 be offered (e.g. Carr and Reed 1993; DeMartini –1 1 3 5 7 9 11 1993). In practice, the actual condition of particu- Years of protection lar sites is unknown. For example, the recruitment (b) 12 source or sink status of sites is rarely established, and this is especially so over long time-scales. Y = 1.61e0.16X r 2 = 0.83 Mean biomass (kg/1000m 8 14.4 WHAT ELSE DO MPAs OFFER?

4 Fishing has effects on the marine environment Apo other than direct impacts on target species (Kaiser reserve and Jennings, Chapter 16, this volume). These include effects on species which interact with the 0 –1 1 3 5 7 9 11 target species and are thus secondarily affected, Years of protection and impacts on habitat, where destructive gears such as trawls are used. The latter may have con- Fig. 14.3 Increase in biomass of large predatory fishes sequences for fish and fisheries if the habitat is [19 species of groupers (Epinephelinae), 11 species important in recruitment. of snappers (Lutjanidae), 6 species of emperors MPA management tends to increase abun- (Lethrinidae) and Carangidae] estimated by underwater dances of site-attached species which have been visual census in two Philippine MPAs, showing a curvi- linear pattern consistent with likely patterns of recruit- depleted by fishing. Through this MPAs can play a ment and growth: (a) Sumilon MPA and (b) Apo MPA. role in the conservation of biological diversity, and (Source: from Russ and Alcala 1996, p. 956.) in creating greater species richness, expressed as the total number of species. Other measures of diversity have been reported in many MPAs grounds that the accumulation of biomass takes a (e.g. Cole et al. 1990; Jennings et al. 1995). Such long time, but is quickly dissipated (e.g. Bohnsack increases do not invariably occur, and there are a 1994). number of reasons for this. In the Philippines, From what has been reviewed above, it can be species richness rose in one MPA (Sumilon), which seen that the siting of MPAs will influence the per- was enforced between bouts of intensive fishing, formance of their ecological and fishery functions. but not in another (Apo) (Russ and Alcala 1998a), With the exception of a few fisheries closures, although differences in the biomass of target existing MPAs have been sited almost exclusively species were significant. In both cases, the princi- in relation to recreational diving and habitat con- pal fishery targets contributed greatly to commu- servation according to criteria which are mostly nity biomass but little to overall numerical 306 Chapter 14 abundance. As a consequence species richness was predation (Chapter 14, Volume 1). This may a poor indicator of the presence or absence of fish- explain why prey fishes are no more abundant in ing even though this was intense, albeit somewhat fished sites than in MPAs in the Seychelles non-selective (Russ and Alcala 1998a). Since most (Jennings et al. 1995) and the Philippines (Russ and data on abundance of target species in MPAs are Alcala 1998a), or grounds in Fiji with very high derived by visual means, the sample areas are fishing pressure when the abundance of predators small, and observer presence can affect them is greatly reduced (Polunin and Jennings 1998). (Kulbicki 1998). In reality effects of MPAs on bio- However, there are cases where linkages between logical diversity are poorly known; for example, species are tighter, and predators may play key- species which are rare and those with large home stone roles, such that community effects of fishing ranges will be undersampled. Given such sampling may be reversed in MPAs, as for triggerfish (Bali- constraints, use of a phrase such as ‘local extirpa- stidae) feeding on sea urchins in Kenyan lagoonal tion’, in the context of a small-scale fishery, needs reefs (McClanahan and Shafir 1990). Other impli- to be classed as provisional until the wider distrib- cations of food webs for changes in communities ution of the species is known. The contention that in MPAs are addressed elsewhere (Chapter 14, species extinction is comparatively less common Volume 1). in the sea than on the land tends to be upheld (e.g. Recreational diving is a point of significant McKinney 1998), but there are points of vulnera- growth in tourism and thus a major source of rev- bility. For example, species of large size with slow enue of marine origin around the world (e.g. Dixon growth and low reproductive rates, such as many et al. 1995). Although there seems scarcely to have turtles, large serranids and elasmobranchs, are vul- been any systematic study of the fact, fishes are nerable (Dulvy et al. 2000; Reynolds et al. 2001). undoubtedly a major focus of much diving and Hotspots of endemism are places where MPAs snorkelling, and in some cases they greatly influ- may play a conservation role. Given the mobility ence perceptions of the underwater experience of the larger species often involved, lack of knowl- (e.g. Andersson 1998; Polunin and Williams 1999). edge of their life histories, and competing uses of MPAs, where they increase the abundance or marine habitat, mean protection of whole popula- size of fishes which are of interest to divers, may tions is unlikely, while planning for piecemeal in- therefore help to generate local revenue through clusion of vulnerable stages and habitats will be increased expenditure on diving, accommodation, difficult. entrance fees and other means (e.g. Dixon et al. The role of MPAs in biodiversity conservation 1995). Estimates of the market value of the stand- could be increased by species interactions, such as ing stock of fishery target fish in a Belize MPA those between predators and prey, and among com- (Polunin and Roberts 1993), when compared with petitors for space. Shifts in community structure the direct revenues to MPAs that are intensively have occurred in many intensively exploited used by divers (e.g. Mattson and DeFoor 1985), in- systems in the tropics and in temperate waters, dicate that marine habitat, in this case reef, can be and there is much debate about the exact nature hundreds of times more valuable per unit area to and causes of these changes (e.g. Hughes 1994; tourism than to capture fisheries. A substantial Christensen 1998; Fogarty and Murawski 1998). part of this tourism revenue is unlikely to reach Although food webs and models of carbon fluxes the local communities involved, and is derived are obvious places to look for synoptic overviews from particular dive sites of limited total area. of systems (e.g. Pauly et al. 1998; Pauly and Chris- Conversely, fishing is a more extensive activity tensen, Chapter 10, this volume), in fact they pro- than most diving, but the value of the reef as source vide pictures only of how species are trophically of catch is lower still when the costs of fishing are linked, and may not be good predictors of change considered. On balance, recreational diving has to in community structure where processes such as be a major potential financial benefit to be derived recruitment, control prey organisms rather than from MPAs. Where this is focused on large or rare Marine Protected Areas 307 species in particular, the implications for enforce- for biodiversity (Auster et al. 1996). Further, in ment need to be carefully considered. The evi- large MPAs, bycatches will no longer be discharged dence from reef fishes is that maximum biomass is back into the environment, and any effects this has built up and large fishes are abundant only if there on abundance of scavengers and predators (e.g. is minimal fishing; a small amount of fishing is Hall 1999) will be reversed. likely to lead to rapid decline in biomass (Fig. 14.4) (see also Russ and Alcala 1996). MPAs have additional value to recreational divers such as through the presence of fragile hard 14.5 REALITIES, ADVOCACY corals and gorgonians. These can be destroyed by AND IMPLEMENTATION fishing (e.g. Shaffer et al. 1998). In addition, recov- OF MPAs ery from the effects of damaging gears such as trawls on benthic organisms and on habitat struc- As in so much environmental literature, there is an ture (e.g. Engel and Kvitek 1998; Thrush et al. inclination in many of the general reviews that 1998; Kaiser et al. 2000) can be expected to occur in have been written about MPAs to list only known MPAs, especially where natural disturbance is advantages of MPAs (e.g. Plan Development Team minor (Jennings and Kaiser 1998). This may have 1990; Ballantine 1991, 1999; Roberts and Polunin significant consequences for fishery species and 1991; Russ et al. 1992; Bohnsack 1993). These can be read as promotional statements of benefits that can be expected from MPAs generally. Costs and uncertainties are inherent in all MPAs (e.g. Dugan and Davis 1993; Allison et al. 1998) and many of (a) 16 14 these are expressed in Table 14.5. Conflict be- ) 2 12 tween potential costs and potential benefits can of 10 course be overplayed, but the water being muddied 8 is a vast middle ground between advocacy and sci- 6 4 ence, and consequences of unbalanced judgements Biomass (g/m 2 can be far-reaching, as evinced by the existence of 0 0 50 100 150 200 250 many protected areas on land. It is true that scien- Fishing pressure (persons/km reef front) tific methods of data gathering and inference are not the best means alone to address urgent de- (b) 30 mands of growing human impacts on the marine )

2 25 environment. In developing countries in particu- 20 lar, advocacy for MPAs does not necessarily solve 15 the underlying problems of seemingly hopeless 10 growing social and economic deprivation (e.g. Pol-

Biomass (g/m 5 unin 1990). In developed regions such as the North 0 0 50 100 150 200 250 Sea or the northeast shelf of North America, it is Fishing pressure (persons/km reef front) hard to see how politicians will legislate for even more draconian measures than those already in Fig. 14.4 Biomass assessed by underwater visual place and derived from conventional fisheries sci- census of (a) epinepheline groupers (genera Anyperodon, ence. These measures have led to the extensive Cephalopholis, Epinephelus, Gracila, Plectropomus and closure of fishing-grounds decided on the basis of Variola) and (b) all piscivorous fishes >30cm in length available data and scientific common sense. It is on reefs in 10 Fijian traditional fishing-grounds varying 600-fold in fishing pressure index from very low levels. worth noting that closures have been applied in Error bars are 95% CL. (Source: after Jennings and areas such as the North Sea where potential bene- Polunin 1996a.) fits for particular stocks such as plaice and cod 308 Chapter 14

Table 14.5 Advantages and disadvantages of permanent or long-term MPAs for site-attached fishes, fisheries management, conservation, science and other stakeholder activities.

Advantages/Likelihoods Disadvantages/Uncertainties

Fisheries Fisheries yield maintained or enhanced through larval dispersal Benefits may be only long term Fisheries yield maintained or enhanced through spillover Higher catch rates confined to a small adjacent area Fluctuations in catch potentially reduced Most of ground closed, yield likely to be reduced Chances of recruitment overfishing reduced by maintained Fishing effects increased because only part of stock exploited spawning stock Fisheries management Undisturbed spawning and/or nursery grounds Unlikely to be useful for migratory species Temptation of fishermen to violate laws reduced Incentive for deliberate poaching increased Data collection needs of management potentially reduced More research needed because of uncertainty as to MPA design Surveillance and enforcement simplified Increased poaching: increased need for surveillance and enforcement Resources potentially available for mariculture and restocking Potential vulnerability to recruitment failure Conservation Habitat protected from damaging fishing techniques Use of damaging methods may be increased outside Endangered/vulnerable species protected Protection unlikely to be for whole life cycle Biodiversity may be increased Most species unaffected Science Areas provided for research on unfished systems Uncertain baselines, pollution uncontrolled Community protected, biodiversity maintained or enhanced Recruitment and system-level effects (e.g. reversibility) unsure Other stakeholders Increased diver use Negative impacts of visitor use and tourism development Management concept more easily understood by public Local resistance to MPA: detailed research needed to justify Areas provided for nature-educational purposes Costs of interpretive facilities and access Economic benefits through tourism Failure of some to benefit economically

have been indicated. The example of the plaice box fact is that not only have few MPAs been estab- emphasizes that wisdom and objective advice will lished in many regions, but where many MPAs not guarantee successful management, but mostly have been gazetted, they have often not been they are sensible preconditions for it. Many have properly managed and so have not met their touched on matters of design (e.g. Carr and Reed objectives (e.g. Stanley 1995; McClanahan 1999). 1993; Dugan and Davis 1993; Rowley 1994; Alli- There are many reasons for this, including lack of son et al. 1998; Ballantine 1999), and I do not intend interest, personnel and money, and opposition to elaborate these further here, beyond what is ex- from other stakeholders. This comes particularly plicit from the empirical and other evidence that I from those who are disadvantaged, as are the fish- have presented. Rather I wish to explore the wider ers by loss of ground. Given the patchy evidence for values and implications of the planning and imple- increased yields of many important resource mentation of MPAs. species arising from MPAs, it is difficult to promise MPAs have positive, negative and unknown future fishery benefits. This is so in developing impacts, and it is surprising that MPAs have rarely countries for reasons of rural poverty, but it is also been subject to the environmental impact analyses difficult for small-scale fisheries in developed that are now widely required for most other devel- countries. If the weak evidence for fisheries bene- opments of any size in areas of natural beauty. The fits is considered, there should be little surprise Marine Protected Areas 309 that scepticism exists; it is well justified on easier when the variety and volume of information scientific grounds. are lower (e.g. Caddy 1999). This is commonly the Declines over time in several of the world’s case in the tropics. In more temperate settings, the marine fish stocks (FAO 1995; Hart and Reynolds, social and economic impacts of MPAs and conse- Chapter 1, Volume 1) may constitute a global crisis quences of fishery science, which is considered by in capture fisheries (Roberts 1997a), yet, although some to have failed (e.g. Roberts 1997a), are surely excessive fishing effort has contributed to decline similar, and will need similarly to be addressed by (e.g. Myers and Barrowman 1996), variation in politicians and managers if measures in either case environmental conditions constitutes a source of are to be a success. great uncertainty for the science, for suitable Small MPAs set up in reef habitats for nature political decision making and for effective man- conservation and not for fishery purposes, mostly agement (e.g. Caddy and Gulland 1983; Botsford et in tropical and subtropical areas (Tables 14.1– al. 1997). Fluctuations in recruitment have been 14.3), have provided much of the impetus for studied for over a century at higher latitudes the advocacy of MPAs (Bohnsack 1993; Roberts (Smith 1994; Smith, Chapter 4, this volume), but 1997a). However, the overwhelming evidence is substantial variations in year-class strength are that MPAs should typically be a large portion of a now being recognized in some tropical species (e.g. fishing-ground if it is to benefit fisheries (e.g. Ferreira and Russ 1995), although probably con- DeMartini 1993; Guénette and Pitcher 1999; cealed in most fisheries through aggregations of Lundberg and Jonzén 1999; Sladek Nowlis and species into broad categories. The arguments that Roberts 1999). Some of the evidence from site- have been presented with respect to the maximiza- attached species on reefs suggests that increased tion of spillover and recruitment effects (Table yields can occur as a result of small MPAs (Table 14.4) highlight the fact that for MPA-orientated 14.3). Yet the impacts which large-scale oceanic fisheries management to be rational, more science events can have on major fisheries (Hofmann and is required, just as modern fisheries management Powell 1998) highlight the vulnerability of small has been underpinned by vigorous investigation MPAs, and the findings and inferences from and reasoned debate (e.g. Smith 1994; Smith, studies of site-attached species, mostly at low lati- Chapter 4, this volume). It is ironic that MPAs tude, and of temperate demersal species, contrast should be confidently advocated as fisheries man- markedly with each other. There are several likely agement measures on the basis of failures in con- reasons for this, including differences in ecology, ventional fisheries science, when the science and different perceptions of scientists studying underpinning MPAs is in its infancy. This is best il- them. It is hard to escape the conclusion that lustrated by the example of recruitment to and MPAs will not everywhere do the same job. They from MPAs, which can clearly be crucial to will need therefore to be designed differently and achievement of management objectives (e.g. Carr appropriately, but further research is needed to and Reed 1993; Allison et al. 1998; Planes et al. predict the outcomes. 2000). What is required is a clearer picture of how The uncertainties of fisheries outcomes from MPAs relate to the wider panoply of fishery man- MPA-orientated management have surely to be agement measures, but one difficulty is that, at treated with openness if such management is to be least in developed countries, the MPA proponents widely adopted. One means towards open dia- are very different from those responsible for fish- logue, particularly with those who stand most to eries management. This may not always be so, but be disadvantaged by large-scale area closure, is it is worth noting that, in contrast, many less-de- through participation by fishers in ‘comanage- veloped countries such as Malaysia, have fisheries ment’ (Hart and Reynolds, Chapter 1, this volume). and MPA management under the same govern- One argument in favour of comanagement mental roof, which may facilitate communica- derives from the contention that in many parts tion, although interdisciplinary work is no doubt of the world local communities have traditionally 310 Chapter 14 managed their resources in areas over which they 20 Votua Tavua have tenure (Johannes 1978; Ruddle et al. 1992). Sophisticated tenure does not go hand in hand 16 with a conservation ethic (e.g. Carrier 1987) and Kubuna the abundant evidence is that such ownership 12 arose for reasons of conflict resolution rather than Verata of resource restraint (e.g. Polunin 1984). The adapt- 8 Ba ability of traditional management to modern man- Management index Levuka agement is thus in question, but it should be the 4 case that local users have potential roles to play in Vitogo resource monitoring and management, and there 0 is some evidence that this is so. For example, 0 0.2 0.4 0.6 0.8 1 in Chile, ‘management and exploitation areas’ Number of licences per unit reef area (km–2) owned by local communities have been beneficial in a mollusc fishery (Castilla and Fernandez 1998). Fig. 14.5 Traditional fishing-grounds (qoliqoli) of Fiji: This case, together with work in Samoa, highlights variations in management regime (‘yes’ or ‘no’ responses to 26 questions relating to the structure of management, the potential uses of cooperative arrangements for the marshalling of information for management, ap- knowledge building and promotion of MPAs (e.g. proach to goodwill payments for licensing, management Castilla 1999; King and Faasili 1999). Local man- measures taken or contemplated, and patrolling and agement on its own would seem typically to be un- enforcement) among the seven qoliqoli which varied in wise for a number of reasons. From work in the access pressure (number of licences issued per ground). Philippines, it is evident that local institutions (Source: Cooke et al. 2000.) may be politically vulnerable (e.g. Russ and Alcala 1999), while experience in Fiji indicates that with- out help local institutions may not be good at dis- must surely increase the likelihood of a wealthy cerning significant changes in the status of their diver returning to a site and of others visiting and resources (Fig. 14.5). Further, the Chilean case enjoying the environment in turn. And in fact it study points to looming conflict between the sus- would appear that in many regions, such as the tainability of resources and wider management ob- Caribbean and Southeast Asia, MPAs have been jectives (e.g. Castilla and Fernandez 1998). Rather, set up with tourism in mind. The most obvious the essence of comanagement is that there are thing that MPAs will do in this respect is to some inputs which are best provided by central au- increase the abundance of site-attached species thorities, and others which are most appropriately which are vulnerable to fishing, and these are an delivered by the local group concerned. important part of what divers and snorkellers In the field of nature conservation, where the come to see (Williams and Polunin 2000). Fishes of concern is to protect biodiversity, vulnerable large size are, however, especially vulnerable to organisms and fragile habitats, the concept of fishing, and these valuable attributes of dive sites MPAs enters different territory. One reason for this need to be conserved, by employing a strict protec- is that the financial value of fragile habitats such as tion regime. It is also probably true that just the reefs can be so much greater on the basis of tourism hint that MPAs are close, or closer, to the natural than that of fisheries. This linkage with tourism of state will be sufficient for many ecotourists to visit course brings a danger of visitor-related damage. and dive in them. In relation to biological diversity, Then, revenue is not the only benefit to consider, the openness of local marine populations and thus although it is an important outcome recognized by their susceptibility to recruitment variability em- politicians and welcomed by the poor. It is evident phasize that the future of biotic assemblages is not that MPAs can contribute much towards such best assured by small MPAs. Either much larger income: well-managed visits to pleasing seascapes areas are needed or networks of areas are required Marine Protected Areas 311 to do the job, but the underlying science needed is organisms and biodiversity as a whole might then poor (e.g. Carr and Reed 1993; Allison et al. 1998). outweigh the lingering doubts about retaining the capacity to exploit the oceans in ways that in prin- ciple have not changed for centuries. In the case of 14.6 CONCLUSIONS some, such as trawling, there is much waste by crude methods. It could happen: liberalization and In the marine environment, human exploitation is globalization of markets have increased, allowing ahead of its time; the environment is different and some to embrace consumption of marine products management is relatively new (Allison et al. 1998). based entirely on imports. As aquaculture shows Fisheries management science has been con- signs of taking off, it could perhaps replace the ducted under conditions that are unfavourable to supply of marine products from capture fisheries, sound science (e.g. Hilborn and Walters 1992; albeit with environmental consequences of its Caddy 1999). Fisheries management science may own. But caution dictates that options should be be derided for the ‘global fisheries crisis’ (Roberts kept open, and, in the meantime, it seems wise to 1997a), but other uncertainties adversely affect the consider MPAs as one of a range of measures which uptake of scientific advice by management (e.g. are available for management of the seas, and Hall 1999), and it would be surprising if an unso- tourism has to feature strongly in that considera- phisticated and inflexible management concept, tion. Apart from reef fish and fisheries, the over- such as represented by MPAs, were to supersede whelming evidence is that management has to be fisheries management with its more detailed sci- large scale, and this implies much more sophisti- entific underpinnings. An important question for cated planning and implementation structures the advocates of MPAs is how the gap can honestly than those required for MPAs. be bridged between the small areas protected (Fig. 14.1) and the large areas required for the delivery of benefits from MPAs to fisheries, ecotourism and ACKNOWLEDGEMENTS biodiversity. In spite of the socioeconomic circum- stances, the spatial scale involved is one which The research and other field experience which un- might more easily be spanned in developing coun- derlie my writing of this chapter have been funded tries than in developed nations. In many develop- from several sources, most notably the Depart- ing countries, the evidence for economic benefits ment for International Development for the from MPAs is greater, and in some cases it seems benefit of developing countries and the Natural that the government structures are less resistant Environment Research Council of the UK. I to the type of interdisciplinary thinking required. especially thank Simon Jennings, Nick Dulvy and Benefits of ecotourism are widely recognized while Adriaan Rijnsdorp for their comments. plentiful relevant resources remain, and lesser sophistication in scientific knowledge probably facilitates discussion at an interdisciplinary REFERENCES level. In contrast, the evidence for benefits being derived from MPAs in developed countries is Alcala, A.C. (1988) Effects of marine reserves on coral minor, government structures seem cumbersome, fish abundances and yields of Philippine coral reefs. and there is a widespread perception that the re- Ambio 17, 194–9. sources are in many cases unredeemably tainted. Alcala, A.C. and Russ, G.R. (1990) A direct test of the The big unknown in all this is, of course, public effects of protective management on abundance and yield of tropical marine resources. Journal du Conseil opinion. If the public voice becomes loud enough, International pour Exploration de la Mer 46, 40–7. then that of the rump fishing community typical Allison, G.W., Lubchenco, J. and Carr, M.H. (1998) of developed countries may scarcely be heard. The Marine reserves are necessary but not sufficient for ma- public perception of the urgency of conserving rare rine conservation. Ecological Applications 8, S79–S92. 312 Chapter 14

Andersson, J. (1998) The value of coral reefs for the cur- Proceedings of International Workshop on Marine rent and potential tourism industry on Unguja. In: Conservation for New Millennium: Marine Protected Francis Johnstone and C. Muhando (eds) Coral Reefs: Areas. Seoul: Korean Ocean Research and Develop- Values, Threats and Solutions, Proceedings of the Na- ment Institute, pp. 3–35. tional Conference on Coral Reefs, Zanzibar Nairobi: Beinssen, K. (1989a) Results of the Boult Reef replenish- UNEP, pp. 81–90. ment area study. Unpublished report, National Parks Appeldoorn, R.S. (1996) Model and method in reef fishery and Wildlife Service, Brisbane, Queensland, Australia. assessment. In: N.V.C. Polunin and C.M. Roberts (eds) Beinssen, K. (1989b) Heron Reef demersal reef fish Reef Fisheries. London: Chapman & Hall, pp. 219– movement study. Unpublished report, Department 48. of Conservation, Parks & Wildlife, Rockhampton, Armstrong, D.A., Wainright, T.C., Jensen, G.C., Dinnel, Queensland, Australia. P.A. and Andersen, H.B. (1993) Taking refuge from by- Beinssen, K. (1990) Heron Reef demersal reef fish move- catch issues: red king crab (Paralithodes camtschati- ment study, update on September 1989 Interim report. cus) and trawl fisheries in the eastern Bering Sea. Unpublished report, Department of Conservation, Canadian Journal of Fisheries and Aquatic Sciences Parks & Wildlife, Rockhampton, Queensland, 50, 1993–2000. Australia. Attwood, C.G. and Bennett, B.A. (1994) Variation in dis- Bell, J.D. (1983) Effects of depth and marine reserve fish- persal of Galjoen (Corachinus capensis) from a marine ing restrictions on the structure of a rocky reef fish as- reserve. Canadian Journal of Fisheries and Aquatic semblage in the north-western Mediterranean Sea. Sciences 51, 1247–57. Journal of Applied Ecology 20, 357–69. Ault, J.S., Bohnsack, J.A. and Meester, G.A. (1998) A Bennett, B.A. and Attwood, C.G. (1991) Evidence for re- retrospective (1979–1996) multispecies assessment covery of a surf-zone fish assemblage following estab- of coral reef fish stocks in the Florida Keys. Fishery lishment of a marine reserve on the southern coast Bulletin 96, 395–414. of South Africa. Marine Ecology Progress Series 75, Auster, P.J., Malatesta, R.J., Langton, R.W., Watling, L., 173–81. Valentine, P.C., Donaldson, C.L.S., Langton, E.W., Bentley, T., Brower, D.J. and Schwab, A.K. (1994) Shepard, A.N. and Babb, I.G. (1996) The impacts of An Introduction to Coastal Zone Management. mobile fishing gear on seafloor habitats in the Gulf of Washington, DC: Island Press. Maine (Northwest Atlantic): implications for conser- Beverton, R.J.H. and Holt, S.J. (1957) On the dynamics of vation of fish populations. Reviews in Fisheries exploited fish populations. Ministry of Agriculture, Science 4, 185–202. Fisheries and Food, Fisheries Investigations (Series 2) Auster, P.J., Watling, L. and Rieser, A. (1997) comment: 19. the interface between fisheries research and habitat Boehlert, G.W. (1996) Larval dispersal and survival in management. North American Journal of Fisheries tropical reef fishes. In: N.V.C. Polunin and C.M. Management 17, 591–5. Roberts (eds) Reef Fisheries. London: Chapman & Hall, Ayling, A.M. and Ayling, A.L. (1986) A biological survey pp. 61–84. of selected reefs in the Capricorn Section of the Great Bohnsack, J.A. (1993) Marine reserves: they enhance fish- Barrier Reef Marine Park. Unpublished report, Great eries, reduce conflicts, and protect resources. Oceanus Barrier Reef Marine Park Authority, Townsville, 36, 63–71. Queensland. Bohnsack, J.A. (1994) How marine fishery reserves can Babcock, R.C., Kelly, S., Shears, N.T., Walker, J.W. and improve reef fisheries. In: Proceedings 43rd Gulf and Willis, T.J. (1999) Changes in community structure in Caribbean Fisheries Institute, Harbor Beach Oceano- temperate marine reserves. Marine Ecology Progress graphic Institution, Fort Pierce, FL, pp. 217–41. Series 189, 125–34. Borley, J.O., Russell, E.S., Graham, M., Wallace, W. and Bailey, R.S. (1991) Changes in the North Sea herring Thursby-Pellam, D.E. (1923) The plaice fishery and the population over a cycle of collapse and recovery. In: war: preliminary report on investigations. Ministry of T. Kawasaki, S. Tanaka, Y. Toba and A. Taniguchi (eds) Agriculture, Fisheries and Food, Fisheries Investiga- Long-term Variability of Pelagic Fish Populations and tions (Series 2) 5. their Environment. Oxford: Pergamon Press, pp. 191–8. Botsford, L.W., Castilla, J.C. and Peterson, C.H. (1997) Ballantine, B. (1991) Marine reserves for New Zealand. The management of fisheries and marine ecosystems. University of Auckland, Leigh Laboratory Bulletin 25, Science 277, 509–15. 196 pp. Bowen, B.K. and Hancock, D.A. (1985) Review of penaeid Ballantine, B. (1999) Marine reserves in New Zealand: fishery management regimes in Australia. In: P.C. the development of the concept and the principles. In: Rothlisberg, B.J. Hill and D.J. Staples (eds) Second Marine Protected Areas 313

Australian National Prawn Seminar. Brisbane: CSIRO, fishes within and among fringing coral reefs in Barba- pp. 247–65. dos. Environmental Biology of Fishes, 57, 11–24. Brothers, E.B., Williams, D.M.B. and Sale, P.F. (1983) Christensen, V. (1998) Fishery-induced changes in a ma- Length of larval life in twelve families of fishes at ‘One rine ecosystem: insight from models of the Gulf of Tree Lagoon’, Great Barrier Reef, Australia. Marine Thailand. Journal of Fish Biology 53 (Supplement A), Biology 76, 319–24. 128–42. Bryant, H.E., Dewey, M.R., Funicelli, N.A., Meineke, Clark, J.R., Causey, B. and Bohnsack, J. (1989) Benefits D.A. and Mengel, J. (1989) Movement of five selected from coral reef protection: Looe Key Reef, Florida. In: species of fish in the Everglades National Park. Bul- O.T. Magoon, H. Converge, D. Miner, L.T. Tobin and letin of Marine Science 44, 515 (abstract). D. Clark (eds) Coastal Zone ’89: Proceedings of the Buxton, C.D. and Allen, J.A. (1989) Mark and recapture Sixth Symposium on Coastal and Ocean Management. studies of tow reef sparids in the Tsitsikamma Coastal New York: American Society of Civil Engineers, pp. National Park. Koedoe 32, 39–45. 3079–86. Buxton, C.D. and Smale, M.J. (1989) Abundance and dis- Cole, R.G., Ayling, T.M. and Creese, R.G. (1990) Effects tribution patterns of three temperate marine reef fish of marine reserve protection at Goat Island, northern (Teleostei: Sparidae) in exploited and unexploited areas New Zealand. NZ Journal of Marine and Freshwater off the southern cape coast. Journal of Applied Ecology Research 24, 197–210. 26, 441–51. Cooke, A.J., Polunin, N.V.C. and Moce, K. (2000) Caddy, J.F. (1999) Fishery management in the twenty- Comparative assessment of stakeholder management first century: will new paradigms apply? Reviews in in traditional Fijian fishing-grounds. Environmental Fish Biology and Fisheries 9, 1–43. Conservation 27, 291–9. Caddy, J.F. and Gulland, J.A. (1983) Historical patterns of Daan, N. (1993) Simulation study of effects of closed fish stocks. Marine Policy 7, 267–78. areas to all fishing, with particular reference to the Carr, M.H. and Reed, D.C. (1993) Conceptual issues rel- North Sea ecosystem. In: K. Sherman, L.M. Alexander evant to marine harvest refuges: examples from tem- and B.D. Gold (eds) Large Marine Ecosystems. perate reef fishes. Canadian Journal of Fisheries and Washington, DC: American Association for the Aquatic Sciences 50, 2019–28. Advancement of Science, pp. 252–8. Carrier, J.G. (1987) Marine tenure and conservation in Davies, C.R. (1996) Inter-reef movement of the Common Papua New Guinea. Problems in interpretation. In: B.J. Coral Trout, Plectropomus leopardus. Unpublished McCay and J.M. Acheson (eds) The Question of the report, Great Barrier Reef Marine Park Authority, Commons. Tucson, AZ: University of Arizona Press, Townsville, Queensland, Australia. pp. 142–67. Davis, G.E. (1977) Effects of recreational harvest on a Castilla, J.C. (1999) Coastal marine communities: trends spiny lobster, Panulirus argus, population. Bulletin of and perspectives from human-exclusion experiments. Marine Science 27, 223–36. Trends in Ecology and Evolution 14, 280–3. Davis, G.E. and Dodrill, J.W. (1980) Marine parks and Castilla, J.C. and Bustamante, R.H. (1989) Human sanctuaries for spiny lobster fisheries management. exclusion from rocky intertidal of Las Cruces, central Proceedings of the Gulf and Caribbean Fisheries Chile: effects on Durvillea antartica (Phaeophyta, Institute 32, 194–207. Durvilleales). Marine Ecology Progress Series 50, Davis, G.E. and Dodrill, J.W. (1989) Recreational fishery 203–14. and population dynamics of spiny lobster, Panulirus Castilla, J.C. and Durán, L.R. (1985) Human exclusion argus, in Florida Bay, Everglades National Park, from the rocky intertidal zone of central Chile: the 1977–1980. Bulletin of Marine Science 44, 78–88. effects on Concholepas concholepas (Gastropoda). DeMartini, E.E. (1993) Modeling the potential of fishery Oikos 45, 391–9. reserves for managing Pacific coral reef fishes. Fishery Castilla, J.C. and Fernandez, M. (1998) Small-scale ben- Bulletin 91, 414–27. thic fisheries in Chile: on co-management and sustain- Dixon, J.A., Scura, L.F. and van’t Hof, T. (1995) Ecology able use of benthic invertebrates. Ecological and microeconomics as ‘joint products’: the Bonarire Applications 8, S124–S132. Marine Park in the Caribbean. In: C.A. Perrings (ed.) Chapman, M.R. and Kramer, D.L. (1999) Gradients in Biodiversity Conservation. The Hague: Kluwer, pp. coral reef fish density and size across the Barbados Ma- 127–45. rine Reserve boundary: effects of reserve protection Done, T.J. and Reichelt, R.E. (1998) Integrated coastal and habitat characteristics. Marine Ecology Progress zone and fisheries ecosystem management: generic Series 181, 81–96. goals and performance indices. Ecological Applica- Chapman, M.R. and Kramer, D.L. (2000) Movement of tions 8, S110–S118. 314 Chapter 14

Duarte, C.M. and Alcaraz, M. (1989) To produce many Gitschlag, G.R. (1986) Movement of pink shrimp in rela- small or few large eggs: a size-dependent reproductive tion to the Tortugas sanctuary. North American Jour- tactic of fish. Oecologia 80, 401–4. nal of Fisheries Management 6, 328–38. Dugan, J.E. and Davis, G.E. (1993) Applications of marine Grimes, C.B. (1987) Reproductive biology of the refugia to coastal fisheries management. Canadian Lutjanidae: a review. In: J.J. Polovina and S. Ralston Journal of Fisheries and Aquatic Sciences 50, 2029–42. (eds) Tropical Snappers and Groupers: Biology and Dulvy, N.K., Metcalfe, J.D., Glanville, J., Pawson, M.G. Fisheries Management. Boulder, CO: Westview Press, and Reynolds, J.D. (2000) Fishery stability, local ex- pp. 239–94. tinctions, and shifts in community structure in skates. Guénette, S. and Pitcher, T.J. (1999) An age-structured Conservation Biology 14, 283–93. model showing the benefits of marine reserves in Edgar, G.J. and Barrett, N.S. (1999) Effects of the declara- controlling overexploitation. Fisheries Research 39, tion of marine reserves on Tasmanian reef fishes, 295–303. invertebrates and plants. Journal of Experimental Hall, S.J. (1999) The Effects of Fishing on Marine Marine Biology and Ecology 242, 107–44. Ecosystems and Communities. Oxford: Blackwell Elgar, M.A. (1990) Evolutionary compromise between Science. few large and many small eggs: comparative evidence Harvey, E. and Shortis, M. (1996) A system for in teleost fish. Oikos 59, 283–7. stereo–video measurement of subtidal organisms. Engel, J. and Kvitek, R. (1998) Effects of otter trawling on Journal of the Marine Technology Society 29, 10–22. a benthic community in Monterey Bay National Ma- Heslinga, G.A., Orak, O. and Ngiramengior, M. (1984) rine Sanctuary. Conservation Biology 12, 1204–14. Coral reef sancturaies for trochus shells. Marine Fish- Fairweather, P.G. (1991) Statistical power and design re- eries Review 46, 73–80. quirements for environmental monitoring. Australian Hilborn, R. and Walters, C.J. (1992) Quantitative Fish- Journal of Marine and Freshwater Research 42, eries Stock Assessment. London: Chapman & Hall. 555–67. Hixon, M.A. (1996) What is ‘recruitment limitation’ any- FAO (1995) The state of the world fisheries and aquacul- way? Reef Encounter 19, 5–6. ture. FAO Fisheries Circular 884. FAO: Rome. Hockey, P.A.R. and Bosman, A.L. (1986) Man as an inter- Ferreira, B.P. and Russ, G.R. (1995) Population structure tidal predator in Transkei: disturbance, community of the coral trout Plectropomus leopardus on fished convergence and management of a natural food re- and unfished reefs off Townsville, Central Great source. Oikos 46, 3–14. Barrier Reef, Australia. Fishery Bulletin 93, 629–42. Hockey, P.A.R. and Branch, G.M. (1997) Criteria, objec- Fogarty, M.J. and Murawski, S.A. (1998) Large-scale dis- tives and methodology for evaluating marine protected turbance and the structure of marine systems: fishery areas in South Africa. South African Journal of Marine impacts on Georges Bank. Ecological Applications 8, Science 18, 369–83. S6–S22. Hofmann, E.E. and Powell, T.M. (1998) Environmental Francour, P. (1991) The effect of protection level on a variability effects on marien fisheries: four case histo- coastal fish community at Scandola, Corsica. Revue ries. Ecological Applications 8, S23–S32. Ecologie Terre et Vie 46, 65–81. Holland, K.N., Peterson, J.D., Wetherbee, B.M. and Lowe, Fulton, E., Kault, D., Mapstone, B. and Sheaves, M. (1999) C.G. (1993) Movements, distribution and growth rates Spawning season influences on commercial catch of white goatfish (Mulloides flavolineatus) in a fish- rates: computer simulations and Plectropomus leopar- eries conservation zone. Bulletin of Marine Science 52, dus, a case in point. Canadian Journal of Fisheries and 982–92. Aquatic Sciences 56, 1096–1108. Holland, K.N., Lowe, C.G., Bradley, M. and Wetherbee, Funicelli, N.A., Johnson, D.R. and Meineke, D.A. (1988) B.M. (1996) Movements and dispersal patterns of blue Assessment of the effectiveness of an existing fish trevally (Caranx melampygus) in a fisheries conserva- sanctuary within the Kennedy Space Center. Unpub- tion zone. Fisheries Research 25, 279–92. lished report, Marine Fisheries Commission of Florida, Horwood, J.W., Nichols, J.H. and Milligan, S. (1998) Miami, USA. Evaluation of closed areas for fish stock conservation. Garcia, S. and Demetropoulos, A. (1986) Management of Journal of Applied Ecology 35, 893–903. Cyprus Fisheries. FAO Fisheries Technical Paper, No. Houde, E.D. (1987) Fish early life dynamics and recruit- 250. FAO: Rome. ment variability. American Fisheries Society Sympo- Carcía-Rubies, A. and Zabala, M. (1990) Effects of total sium 2, 17–29. fishing prohibition on the rocky fish assemblages of Hughes, T.P. (1994) Catastrophes, phase shifts, and large- Medes Islands marine reserve (NW Mediterranean). scale degradation of a Caribbean coral reef. Science Scientia Marina 54, 317–28. 265, 1547–51. Marine Protected Areas 315

ICES (1990) Report of the ICES Advisory Committee on Kelly, S. (1999) Marine reserves and the spiny lobster Fishery Management. ICES Co-operative Research Jasus edwardsii. PhD thesis, University of Auckland, Report 168, Copenhagen. New Zealand. ICES (1999) Report of the Workshop on the Evalu- King, M. (1995) Fisheries Biology, Assessment and Man- ation of the Plaice Box. International Council for the agement. Oxford: Fishing News Books. Exploration of the Sea, Copenhagen CM 1999/D: King, M. and Faasili, U. (1999) Community-based man- 6. agement of subsistence fisheries in Samoa. Fisheries Jennings, S., Grandcourt, E.M. and Polunin, N.V.C. Management and Ecology 6, 133–44. (1995) The effects of fishing on the diversity, biomass Klima, E.F., Matthews, G.A. and Patella, F.J. (1986) Syn- and trophic structure of Seychelles’ reef fish commu- opsis of the Tortugas pink shrimp fishery, 1960–1983, nities. Coral Reefs 14, 225–35. and the impact of the Tortuga sancturay. North Ameri- Jennings, S. and Kaiser, M.J. (1998) The effects of fishing can Journal of Fisheries Management 6, 301–10. on marine ecosystems. Advances in Marine Biology Kramer, D.L. and Chapman, M.R. (1999) Implications of 34, 201–352. fish home range size and relocation for marine reserve Jennings, S., Kaiser, M.J. and Reynolds, J.D. (2001) Ma- function. Environmental Biology of Fishes 55, 65–79. rine Fisheries Ecology. Oxford: Blackwell. Kulbicki, M. (1998) How acquired fish behaviour of Jennings, S. and Polunin, N.V.C. (1996a) Impacts of commercial reef fish may influence results obtained predator depletion by fishing on the biomass and diver- from visual censuses. Journal of Experimental Marine sity of non-target fish communities. Coral Reefs 16, Biology and Ecology 222, 11–30. 71–82. Lasiak, T. and Dye, A. (1989) The ecology of the brown Jennings, S. and Polunin, N.V.C. (1996b) Fishing strate- mussel Perna perna in Transkei, southern Africa: gies, fishery development and socioeconomics in tra- implications for the management of a traditional food ditionally managed Fijian fishing grounds. Fisheries resource. Biological Conservation 47, 245–57. Management and Ecology 3, 335–47. Lauck, T., Clark, C.W., Mangel, M. and Munro, G.R. Jennings, S., Reynolds, J.D. and Mills, S.C. (1997) Life his- (1998) Implementing the precautionary principle in tory correlates of responses to fisheries exploitation. fisheries management through marine reserves. Eco- Proceedings of the Royal Society of London B 265, logical Applications 8, S72–S78. 333–9. Leary, T.R. (1985) Review of the Gulf of Mexico manage- Jennings, S., Reynolds, J.D. and Polunin N.V.C. (1999) ment plan for Shrimp. In: P.C. Rothlisberg, B.J. Hill and Predicting the vulnerability of tropical reef fishes D.J. Staples (eds) Second Australian National Prawn to exploitation: an approach based on phylogenies Seminar. Brisbane: CSIRO, pp. 267–74. and life histories. Conservation Biology 13, 1466– Lundberg, P. and Jonzén, N. (1999) Spatial population 75. dynamics and the design of marine reserves. Ecology Johannes, R.E. (1978) Traditional marine conservation Letters 2, 129–34. methods in Oceania and their demise. Annual Review Man, A., Law, R. and Polunin, N.V.C. (1995) Role of of Ecology and Systematics 9, 349–64. marine reserves in recruitment to reef fisheries: a Jones, G.P., Milicich, M.J., Emslie, M.J. and Lunow, C. metapopulation model. Biological Conservation 71, (1999) Self-recruitment in a coral reef fish population. 197–204. Nature 402, 802–4. Margetts, A.R. and Holt, S.J. (1948) The effects of the Jones, P.J.S. (1994) A review and analysis of the objectives 1939–1945 war on the English North Sea trawl fish- of marine nature reserves. Ocean and Coastal Man- eries. Rapports du Conseil Permanent International agement 24, 149–78. pour l’Exploration de la Mer 122, 26–46. Jones, P.J.S. (1999) Marine nature reserves in Britain: Mattson, J.S. and DeFoor, J.A. (1985) Natural resource past lessons, current status and future issues. Marine damages: restitution as a mechanism to slow destruc- Policy 23, 375–96. tion of Florida’s natural resources. Journal of Land Use Kaiser, M.J., Spence, F.E. and Hart, P.J.B. (2000) Harvest- Environmental Law 1, 295–319. ing with due care and attention: fishing gear restric- McCay, B.J. (1988) Muddling through the clam beds: co- tions conserve benthic habitat complexity. operative management of New Jersey’s hard clam Conservation Biology 14, 1512–25. spawner sanctuaries. Journal of Shellfish Research 7, Kelleher, G., Bleakley, C. and Wells, S. (1995) Introduc- 327–40. tion. In: G. Kelleher, C. Bleakley and S. Wells (eds) A McClanahan, T.R. (1999) Is there a future for coral reef Global Representative System of Marine Protected parks in poor tropical countries? Coral Reefs 18, 321–5. Areas, Volume I. Washington, DC: The World Bank, McClanahan, T.R. and Kaunda-Arara, B. (1996) Fishery pp. 1–15. recovery in a coral-reef marine park and its effect 316 Chapter 14

on the adjacent fishery. Conservation Biology 10, Hereu, B., Milazzo, M. and Zabala, M. (2000) Tropic 1187–99. cascades in fisheries and protected-area management McClanahan, T.R. and Muthiga, N.A. (1988) Changes in of benthic marine ecosystems. Environmental Con- Kenyan coral reef community structure and function servation 27, 179–200. due to exploitation. Hydrobiologia 166, 269–76. Pipitone C., Badalamenti, F., D’Anna, G. and Patti, B. McClanahan, T.R. and Shafir, S.H. (1990) Causes and (2000) Fish biomass increase after a four-year trawl ban consequences of sea urchin abundance and diversity in in the Gulf of Castellammare (NW Sicily, Mediter- Kenyan coral reef lagoons. Oecologia 83, 362–70. ranean Sea). Fisheries Research 48, 23–30. McCormick, M.I. and Choat, J.H. (1987) Estimating total Plan Development Team (1990) The potential of marine abundance of a large temperate reef fish using visual fishery reserves for reef management in the U.S. South- strip-transects. Marine Biology 96, 469–78. ern Atlantic. NOAA NMFS Technical Memorandum McKinney, M.L. (1998) Is marine biodiversity at less risk? NMFS-SEFC-261, NOAA NMFS South Fisheries Evidence and implications. Diversity and Distribu- Center, Miami, FL. tions 4, 3–8. Planes, S., Galzin, R., Garcia-Rubies, A., Goñi, R., Moreno, C.A., Sutherland, J.P. and Jara, H.F. (1984) Man Harmelin, J.-G., LeDiréach, L., Lenfant, P. and as a predator in the intertidal zone of southern Chile. Quetglas, A. (2000) Effects of marine protected areas Oikos 42, 155–60. on recruitment processes with special reference to Mosquera, I., Côté, I.M., Jennings, S. and Reynolds, J.D. Mediterranean littoral ecosystems. Environmental (2000) Conservation benefits of marine reserves for Conservation 27, 1–18. fish populations. Animal Conservation 4, 321–32. Polacheck, T. (1990) Year round closed areas as a manage- Munro, J.L. (1999) Marine Protected Areas and the Man- ment tool. Natural Resource Modeling 4, 327–54. agement of Coral Reef Fisheries. Unpublished techni- Polunin, N.V.C. (1984) Do traditional marine ‘reserves’ cal report, ICLARM Caribbean/Eastern Pacific Office, conserve? A view of Indonesian and Papua New Tortola, British Virgin Islands. Guinean evidence. In: K. Ruddle and T. Akimichi (eds) Munro, J.L. and Watson, M. (1999) Comparative settle- Maritime Institutions in the Western Pacific, Senri ment and recruitment rates of coral reef fishes in mod- Ethnological Studies 17. Osaka: National Museum of erately exploited and over-exploited Caribbean coral Ethnology, pp. 267–83. reef ecosystems. In: Marine Protected Areas and the Polunin, N.V.C. (1990) Marine regulated areas: an ex- Management of Coral Reef Fisheries. Unpublished panded approach for the tropics. Resource Manage- technical report, ICLARM Caribbean/Eastern Pacific ment and Optimization 7, 283–99. Office, Tortola, British Virgin Islands, pp. 9–14. Polunin, N.V.C. (1997) Fishing effects on target species Myers, R.A. and Barrowman, N.J. (1996) Is fish recruit- and the design of reserves for recovery. In: T.J. Pitcher ment related to spawner abundance? Fishery Bulletin (ed.) The Design and Monitoring of Marine Reserves. 94, 707–24. University of British Columbia, Fisheries Centre Re- OECD (1993) Coastal Zone Management: Integrated search Report 5(1), pp. 8–9. Policies. Paris: Organization for Economic Coopera- Polunin, N.V.C. (1999) Fishery-independent Assessment tion for Development. of Management Effects on Community Fishery Re- Parrish, J.H. (1987) The trophic biology of snappers and sources in Fiji and Vanuatu. Department of Marine groupers. In: J.J. Polovina and S. Ralston (eds) Tropical Sciences, University of Newcastle, UK, Marine Sys- Snappers and Groupers: Biology and Fisheries Man- tems Research Group Technical Report 6. agement. Boulder, CO: Westview Press, pp. 405–63. Polunin, N.V.C. and Jennings, S. (1998) Differential ef- Pastoors, M.A., Rijnsdorp, A.D. and Van Beek, F.A. (2000) fects of small-scale fishing on predatory and prey fishes Effects of a partially closed area in the North Sea on Fijian reefs. In: D.M. Newbery, H.H.T. Prins and N. (‘plaice box’) on stock development of plaice. ICES Brown (eds) Dynamics of Tropical Communities, Journal of Marine Science 57, 1014–22. British Ecological Society Symposium 37. Oxford: Pauly, D., Christensen, V., Dalsgaard, J., Froese, R. and Blackwell Science, pp. 95–124. Torres, F. (1998) Fishing down marine food webs. Polunin, N.V.C. and Roberts, C.M. (1993) Greater bio- Science 279, 860–3. mass and value of target coral-reef fishes in two small Piet, G.J. and Rijnsdorp, A.D. (1998) Changes in the dem- Caribbean marine reserves. Marine Ecology Progress ersal fish assemblage in the southeastern North Sea Series 100, 167–76. following the establishment of a protected area (‘plaice Polunin, N.V.C. and Williams, I.D. (1999) Ecological Im- box’). ICES Journal of Marine Science 55, 420–9. pacts of Caribbean Marine Reserves. Department of Pinnegar, J.K., Polunin, N.V.C., Francour, P., Marine Sciences, University of Newcastle, UK, Ma- Badalamenti, F., Chemello, R., Harmelin-Vivien, M-L., rine Systems Research Group Technical Report 5. Marine Protected Areas 317

Quinn, J.F., Wing, S.R. and Botsford, L.W. (1993) Harvest of large predatory fish. Ecological Applications 6, refugia in marine invertebrate fisheries: models and 947–61. applications to red sea urchin, Strongylocentrotus Russ, G.R. and Alcala, A.C. (1998a). Natural fishing ex- franciscanus. American Zoologist 33, 537–50. periments in marine reserves 1983–1993: community Rakitin, A. and Kramer, D.L. (1996) Effect of a marine re- and trophic responses. Coral Reefs 17, 383–97. serve on the distribution of coral reef fishes in Barba- Russ, G.R. and Alcala, A.C. (1998b) Natural fishing ex- dos. Marine Ecology Progress Series 131, 97–113. periments in marine reserves 1983–1993: roles of life Reynolds, J.D., Jennings, S. and Dulvy, N.K. (2001). Life history and fishing intensity in family responses. histories of fishes and population responses to ex- Coral Reefs 17, 399–416. ploitation. In: J.D. Reynolds, G.M. Mace, K.H. Redford Russ, G.R. and Alcala, A.C. (1999) Management histories and J.G. Robinson (eds.). Conservation of Exploited of Sumilon and Apo Marine Reserves, Philippines, and Species. Cambridge: Cambridge University Press. their influence on national marine resource policy. Rice, M.A., Hickox, C. and Zehra, I. (1989) Effects of in- Coral Reefs 18, 307–19. tensive fishing effort on the population structure of Russ, G.R., Alcala, A.C. and Cabanban, A.S. (1992) Ma- quahogs, Mercenaria mercenaria (Linnaeus, 1758), in rine reserves and fisheries management on coral reefs Narragansett Bay. Journal of Shellfish Research 8, with preliminary modelling of the effects on yield per 345–54. recruit. In: Proceedings of the 7th International Coral Rijnsdorp, A.D. and Pastoors, M.A. (1995) Modelling the Reef Symposium, Guam 2, 978–85. spatial dynamics and fisheries of North Sea plaice Rutherford, E.S., Tilmant, J.T., Thue, E.B. and Schmidt, (Pleuronectes platessa L.) based on tagging data. ICES T.W. (1989) Fishery harvest and population dynamics Journal of Marine Science 52, 963–80. of grey snapper, Lutjanus griseus, in Florida Bay and ad- Rijnsdorp, A.D., Van Beek, F.A., Flatman, S., Millner, jacent waters. Bulletin of Marine Science 44, 139–54. R.M., Riley, J.D., Giret, M. and De Clerck, R. (1992) Re- Sadovy, Y.J. (1996) Reproduction of reef fishery species. cruitment of sole stocks, Solea solea (L.), in the North- In: N.V.C. Polunin and C.M. Roberts (eds) Reef Fish- east Atlantic. Netherlands Journal of Sea Research 29, eries. London: Chapman & Hall, pp. 15–59. 173–92. Sale, P.F. (1991) Reef fish communities: open nonequilib- Roberts, C.M. (1997a) Ecological advice for the global rial systems. In: P.F. Sale (ed.) The Ecology of Fishes on fisheries crisis. Trends in Ecology and Evolution 12, Coral Reefs. San Diego: Academic Press, pp. 564–98. 35–8. Samoilys, M.A. (1988) Abundance and species richness of Roberts, C.M. (1997b) Connectivity and management of coral reef fish on the Kenyan coast: the effects of pro- Caribbean coral reefs. Science 278, 1454–7. tective management and fishing. In: Proceedings of the Roberts, C.M. and Polunin, N.V.C. (1991) Are marine 6th International Coral Reef Symposium 2, 261–6. reserves effective in management of reef fisheries? Sánchez Lizaso, J.L., Goñi, R., Reñones, O., García- Reviews in Fish Biology and Fisheries 1, 65–91. Charton, J.A., Galzin, R., Bayle, J.T., Sánchez-Jerez, P., Roberts, C.M. and Polunin, N.V.C. (1992) Effects of ma- Pérez Ruzafa, A. and Ramos, A.A. (2000) Density rine reserve protection on northern Red Sea fish popu- dependence in marine protected populations: a review. laitons. In: Proceedings of the 7th International Coral Environmental Conservation 27, 144–58. Reef Symposium University of Guam 2, 969–77. Shaffer, C.S., Inglis, G.L., Johnson, V.Y. and Marshall, Roberts, T.W. (1986) Abundance and distribution of N.A. (1998) Visitor experiences and perceived condi- pink shrimp in and around the Tortugas Sanctuary, tions on day trips to the Great Barrier Reef. CRC 1981–1983. North American Journal of Fisheries Reef Research Technical Report 21, Townsville, Management 6, 311–27. Queensland, Australia. Rowley, R.J. (1994) Marine reserves in fisheries manage- Shepherd, S.A. (1990) Studies on Southern Australian ment. Aquatic Conservation: Marine and Freshwater abalone (genus Haliotis). XII. Long-term recruitment Ecosystems 4, 233–54. and mortality dynamics of an unfished population. Ruddle, K., Hviding, E. and Johannes, R.E. (1992) Marine Australian Journal of Marine and Freshwater Re- customary management in the context of customary search 41, 475–92. tenure. Marine Resource Economics 7, 249–73. Sladek Nowlis, J.S. and Roberts, C.M. (1999) Fisheries Russ, G.R. (1985) Effects of protective management on benefits and optimal design of marine reserves. Fishery coral reef fishes in the central Philippines. In: Proceed- Bulletin 97, 604–16. ings of the 5th International Coral Reef Congress Smith, A.H. and Berkes, F. (1991) Solutions to the Tahiti, 4, 219–24. ‘Tragedy of the Commons’: sea-urchin management in Russ, G.R. and Alcala, A.C. (1996) Marine reserves: rates St Lucia, West Indies. Environmental Conservation and patterns of recovery and decline in abundance 18, 131–6. 318 Chapter 14

Smith, T.D. (1994) Scaling Fisheries. Cambridge: at Barbados, West Indies. Environmental Biology of Cambridge University Press. Fishes 55, 53–63. Stanley, S. (1995) Wider Caribbean. In: G. Kelleher, C. Wallace, S.S. (1998) Evaluating the effects of three forms Bleakley and S. Wells (eds) A Global Representative of marine reserve on northern abalone populations in System of Marine Protected Areas, Volume II. British Columbia, Canada. Conservation Biology 13, Washington DC: The World Bank, pp. 13–41. 882–7. Stoner, A.W. and Ray, M. (1996) Queen conch, Wantiez, L., Thollot, P. and Kulbicki, M. (1997) Effects of Strombus gigas, in fished and unfished locations of the marine reserves on coral reef fish communities from Bahamas: effects of a marine fishery reserve on adults, five islands in New Caledonia. Coral Reefs 16, 215–24. juveniles, and larval production. Fishery Bulletin 94, Watson, M. (1996) The role of protected areas in the man- 551–65. agement of Kenyan reef fish stocks. PhD thesis, Swearer, S.E., Caselle, J.E., Lea, D.W. and Warner, R.R. University of York, England. (1999) Larval retention and recruitment in an Weil, M.E. and Laughlin, R.G. (1984) Biology, population island population of a coral-reef fish. Nature 402, 799– dynamics and reproduction of the queen conch Strom- 802. bus gigas Linné in the Archipelago de los Roques Tegner, M.J. (1989) The California abalone fishery: pro- National Park. Journal of Shellfish Research 4, 45–62. duction, ecological interactions, and prospects for the White, A.T. (1988) The effect of community-managed future. In: J.F. Caddy (ed.) Marine Invertebrate Fish- marine reserves in the Philippines on their associated eries: their Assessment and Management. Oxford: coral reef fish populations. Asian Fisheries Science 2, Blackwell Science, pp. 401–20. 27–41. Tegner, M.J. (1992) Brood-stock transplants as an Williams, I.D. and Polunin, N.V.C. (2000) Differences approach to abalone stock enhancement. In: S.A. between protected and unprotected Caribbean reefs in Shepherd, M.J. Tegner and S.A. Guzman del Proo attributes preferred by dive tourists. Environmental (eds) Abalone of the World: Biology, Fisheries and Conservation 27, 382–91. Culture. Oxford: Blackwell Science, pp. 461–73. Willis, T.J., Millar, R.B. and Babcock, R.C. (2000) Detec- Thresher, R.E. and Brothers, E.B. (1989) Evidence of intra- tion of spatial variability in relative density of fishes: and inter-oceanic regional differences in the early life comparison of visual census, angling, and baited history of reef-associated fishes. Marine Ecology underwater video. Marine Ecology Progress Series 198, Progress Series 57, 187–205. 249–60. Thrush, S.F., Hewitt, J.E., Cummings, V.J., Dayton, P.K., Wootton, R.J. (1990) Ecology of Teleost Fishes. London: Cryer, M., Turner, S.J., Funnell, G.A., Budd, R.G., Chapman & Hall. Milburn, C.J. and Wilkinson, M.R. (1998) Disturbance Yamasaki, A. and Kuwahara, A. (1990) Preserved area to of the marine benthic habitat by commercial fishing: effect recovery of overfished Zuwai crab stocks off impacts at the scale of the fishery. Ecological Applica- Kyoto Prefecture. In: Proceedings of the International tions 8, 866–79. Symposium on King and Tanner Crabs, Alaska Sea Tupper, M. and Juanes, F. (1999) Effects of a marine Grant College Program. Fairbanks, Alaska: University reserve on recruitment of grunts (Pisces: Haemulidae) of Alaska, pp. 575–85. 15 Exploitation and other Threats to Fish Conservation

JOHN D. REYNOLDS, NICHOLAS K. DULVY AND CALLUM M. ROBERTS

15.1 INTRODUCTION at stake, politics has played a large role in the uptake of management advice, often erring on The traditional goal in fisheries management has the side of continued employment rather than been to obtain continuing yields from a living ‘re- more stringent, less socially palatable manage- source’. Concerns are raised when populations ment measures (Hart and Reynolds, Chapter 1, fall below levels that provide adequate yields or this volume). which fail to meet other specified reference points Until recently, such failures of fisheries man- (Shepherd and Pope, Chapter 8, this volume). Un- agement policy have remained outside the fortunately, fishers and fisheries biologists have mainstream conservation movement. Indeed, had a lot to be concerned about lately, as maximum conservationists have generally taken much less sustainable yields have been exceeded for many interest in marine and freshwater environments fisheries (Jennings et al. 2001b) with many stocks than in terrestrial habitats. For example, a survey now in decline. There have been some spectacular of papers published in the journal Conservation declines of species with a wide spectrum of life Biology found that only 5% were for marine histories and habitats, including various stocks species and habitats, 9% were freshwater, and 67% of Atlantic cod, Gadus morhua, Peruvian an- were terrestrial (Irish and Norse 1996). While there choveta, Engraulis ringens, southern bluefin have been some notable public concerns about tuna, Thunnus maccoyii, swordfish, Xiphius selected issues such as whaling, declines in fish gladius, and sablefish, Anoplopoma fimbrae populations have dwelt in the domain of ‘manage- (Fig. 15.1). ment failures’ rather than ‘conservation problems’ The economic and social hardships caused by (Reynolds and Jennings 2000). This inattention population declines in so many fisheries around to conservation issues in fishing has started to the world have received a great deal of deserved at- change during the past decade as conservationists tention. The poorest of the world’s countries (with have begun to worry about the possibility that fish a per capita gross domestic product

(a) Northern cod (b) North Sea cod 1600 300 250 1200 200

800 150

100

(thousand tonnes) 400 (thousand tonnes)

Spawning stock biomass 50 Spawning stock biomass 0 0 1962 1967 1972 1977 1982 1987 1992 1960 1965 1970 1975 1980 1985 1990 1995 2000 Year Year

(c) Peruvian anchoveta (d) Southern bluefin tuna 15 6000 ) 3 5000 10 × 10 4000

3000 Landings 5 2000 (million tonnes)

1000 Number of individuals ( 0 0 1950 1960 1970 1980 1990 2000 1965 1975 1985 Year Year

(e) Swordfish (f) Sablefish 180 40 160 35 140 30 120 25 100 20 80 15 60 (thousand tonnes) (thousand tonnes) 10 40 Spawning stock biomass 5 Female spawning biomass 20 0 0 1975 1980 1985 1990 1995 2000 1965 1970 1975 1980 1985 1990 1995 2000 Year Year

Fig. 15.1 Fish stock declines: (a) northern cod, Gadus morhua (Hutchings and Myers 1994); (b) North Sea cod, Gadus morhua (ICES 2001); (c) Peruvian anchoveta, Engraulis ringens (FAO 1999); (d) southern bluefin tuna, Thunnus maccoyii (Matsuda et al. 1998); (e) swordfish, Xiphius gladius; (f) sablefish, Anoplopoma fimbrae (http://www.mscs.dal.ca/~myers/welcome.html). Conservation and Exploitation 321 ment during the 1970s and 1980s, fishes are 9% now featuring alongside other taxa as the public and scientists ask questions about links between exploitation and conservation (e.g. Mace and 17% 26% Hudson 1999; Hutchings 2001a; Reynolds et al. 2001a). The goal of this chapter is to review the role of exploitation in causing conservation problems for fish species. We will review the evidence for severe declines in freshwater and marine fish populations which could lead to extirpation or extinction. We examine biological attributes of fishes and socio- economic aspects of fisheries that render species vulnerable. This leads to a consideration of how ‘conservation’ of exploited species means different things to different people, as shown by difficulties in assessing the threatened status of marine fishes. We hope that this discussion will help to bridge 48% the scientific gap between different approaches to ‘conservation’ of exploited species. Underexploited Overexploited 15.2 GLOBAL STATUS Fully exploited Depleted OF EXPLOITED FISH POPULATIONS Fig. 15.2 Percentages of the world’s fish stocks that range from underexploited to depleted. (Source: from FAO 1999.) The fallacy that there are always more fish in the sea has officially ended. Marine capture fisheries produced 86 million tonnes in 1998, valued at 15.3 EXTINCTION US$76 billion. The global marine fish catch was 15.3.1 Recent extinctions thought to have levelled out in the 1990s, but a pre- dictive catch model suggests the global marine fish There are severe problems in estimating how catch peaked in 1988 at 78 million tonnes and has many fish species have become extinct because of since declined to 69 million tonnes (Watson and the difficulties of sampling aquatic habitats suffi- Pauly 2001; see Hart and Reynolds, Chapter 1, Vol- ciently well to be sure that a fish has truly disap- ume 1). These figures omit the highly variable peared from its entire range (Smith et al. 1993; Peruvian achoveta catch and correct for massive Carlton et al. 1999; McKinney 1999). It is no longer misreporting by China (Watson and Pauly 2001). fashionable for funding agencies to pay for basic The FAO report that approximately half of the taxonomy and collecting trips. Thus, lack of major fish stocks are fully exploited and very sampling effort as well as taxonomic uncertainties close to their maximum sustainable limits, with raise real difficulties in assessing accurately another quarter overexploited or depleted (Fig. whether or not species have become extinct. 15.2). Only the remaining quarter of the world’s These problems are illustrated by an attempt to fish stocks are considered to be under- or moder- convey the strength of evidence for extinction of ately exploited (FAO 1999). freshwater fishes, which has led to ongoing refine- ments of criteria and a database showing the level 322 Chapter 15 of support for apparent recent (since ad 1500) ex- viridis) was described from shallow waters tinctions of fishes (Harrison and Stiassny 1999; around the island of Mauritius in 1839 but has not http://creo.amnh.org/). Examples of criteria that been seen since. It may have suffered from degrada- have been proposed include that the species’ name tion of reefs in the area, due to sedimentation should be taxonomically valid, and that attempts and nutrient pollution. Similarly, the deepwater to collect the species during appropriate surveys angelfish, Apolemichthys guezi, endemic to the have failed. Both of these requirements are often nearby island of La Réunion, seems also to have sticking points in conservation assessments of disappeared, though further sampling will be aquatic species. There has also been debate about necessary to confirm this. Morris et al. (2000) also whether one should wait some arbitrary period noted that several species of tropical groupers have of time (e.g. 50 years) to be sure that a species has not been seen for long periods since their first truly gone, but this requirement has been dropped description. Although general rarity and poor by the World Conservation Union (IUCN’s) Red sampling are probably to blame for their ‘disap- List rules, as well as by the American Museum of pearance’, groupers are highly vulnerable to over- Natural History’s Committee on Recently Extinct fishing (Coleman et al. 1996). Thus, while some Organisms (http://creo.amnh.org/). of these marine taxa may rise from the dead, ex- Table 15.1 shows 34 fish species whose taxo- tinction of highly localized species in areas subject nomic status is clear and for which surveys have to detrimental human activities cannot be ruled been adequate to be reasonably certain that they out. are extinct. This list does not include Lake Victoria cichlids, largely because of inadequate surveys 15.3.2 Palaeoextinctions (Harrison and Stiassny 1999). When these species are included, along with other cases of unresolved The fossil record may shed some light on whether extinctions, the database includes a total of 164 extinction rates in marine habitats are really lower species. than in non-marine habitats. McKinney (1998) It is noteworthy that all known cases of recent showed that in a variety of non-fish taxa, marine extinctions of fishes (since ad 1500) have hap- lineages have persisted on average for five times pened to be freshwater species, despite the fact that longer in the fossil record than have non-marine marine taxa account for roughly 60% of described lineages. However, it is not clear whether these dif- species (Gill and Mooi, Chapter 2, Volume 1). ferences are due to differences among habitats While the compilation by Harrison and Stiassny per se, or due to differences among the taxa them- (1999) focused on freshwater taxa, our own survey selves, since none of the taxa analysed had repre- of the literature and discussions with colleagues sentatives in both environments. If these broad did not reveal any marine species that we could add taxonomic comparisons also apply to fishes, they to the list with complete confidence. However, would support the impression of lower rates of there are at least three reasonable candidates that recent extinctions in marine habitats. may well be extinct (Roberts and Hawkins 1999; McKinney (1998) also looked at how duration of Hawkins et al. 2000). The Galapagos damselfish species in the fossil record related to the fraction of (Azurina eupalama) is a planktivorous species a taxon that was listed as threatened by the World which disappeared during the 1982–3 El Niño Conservation Union (IUCN 1996). Fishes repre- in the eastern Pacific, one of the most intense El sented one data point among a number of broad Niños for the last several hundred years (Glynn taxonomic groupings. There was a positive rela- 1988). The warming associated with the El Niño tionship between extinction rates in the geologic shut down the upwellings that fuelled plankton past and the percentage of species currently threat- production for nearly one year and the Galapagos ened. Of course, these data are confounded because damselfish has not been seen since, despite the most threatened taxa also tend to be the largest thorough surveys. The green wrasse (Anampses and best known, and the best-known species do Conservation and Exploitation 323 her species, 19031988198419031937 HM – P 1889 HM 1959 HM HM – P 1985 IN 1880 HM – IN 1948 HM – P 19681923 P – IN D 1952 HM – P 1930 HM – IN HM – IN HM – IN D IN – OH HY HM – IN OE 1983194419271921 P – IN D 1881 HM – P 1977 HM 19571954 HM – IN OE 1957 P – OE 1957 HD 1938 HM – P HD 1922 HM – P HD 1969 HM – P HD 1961 HM – IN 1924 IN 1921 HM – IN OE 1970 HM – P IN 1940 HM – P IN 1922 HM – IN OE HM – IN OE HM – P IN HM HM – IN OE > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ).) http://creo.amnh.org/ Philippines Yugoslavia China Philippines stickleback stumptooth minnowstumptooth Mexico pupfishcachorrito del la PresaParras pupfish Titicaca orestiasLake Mexico whiteline topminnow ninespineKyoyto Bolivia, Peru States United Mexico Mexico bunguPoso Parras characodon killifishAsh Meadows Japan Amistad gambusia Zealand graylingNew Mexico States United ciscoedeepwater Indonesia troutsilver States United Zealand New States Canada, United States United duck-billed buntingiduck-billed ginger pearlfish sucker River Snake suckerharelip Indonesia Brazil States United Mexican daceplateau dace States United endorheic dacethicktail chub spinedacepahranagat bagangan Mexico Ameca shiner Mexico States United Mexico Durango shiner 1893 States United bitungu splittailClear Lake daceLas Vegas Philippines Mexico Mexico palata HM States United States United Philippines Philippines CyprinodonCyprinodonCyprinodonOrestiasFundulus ceciliae Pungitius inmemoriam latifasciatus WeberogobiusCharacodon cuvieri albolineatus Empetrichthys kaibarae Gambusia amadi Prototroctes garmani Coregonus merriami Salvelinus amistadensis oxyrhynchus johannae agassizi LeptolebiasChasmistesMoxostomaCephalakompsusChondrostoma marmoratus Cyprinus muriei pachycheilus Evarra lacerum Evarra scodrensis EvarraGilaLepidomeda yilongensis Mandibularca bustamantei Notropis eigenmanni Notropis tlahuacensis Ospatulus altivelis resinus Ospatulus crassicauda PogonichthysRhinichthys amecae Sprattelicypris aulidion palaemophagus Stypodon trunculatus ciscoides deaconi palata signifer Adrianichthys kruyti Species of fish for which there is strong evidence of extinction. This list does not include Lake Victoria cichlids, nor many ot Species of fish for which there is strong evidence extinction. This list does not include Lake Victoria : hybridization; IN, introduced species; Codes for cause of extinction: HD, habitat destruction; HM, habitat modification; HY, Cyprinodontidae Cyprinodontidae Cyprinodontidae Cyprinodontidae Fundulidae Gastoresteidae Gobiidae Goodeidae Goodeidae Poeciliidae Retropinnidae Salmonidae Salmonidae Aplocheilidae Catostomidae Catostomidae Cyprinidae Cyprinidae Cyprinidae Cyprinidae Cyprinidae Cyprinidae Cyprinidae Cyprinidae Cyprinidae Cyprinidae Cyprinidae Cyprinidae Cyprinidae Cyprinidae Cyprinidae Cyprinidae Cyprinidae Table 15.1 Table due to uncertainties about taxonomic status or incomplete surveys of appropriate habitat. FamilyAdrianichthyidae Genus Species Common name Former distribution Extinction year Extinction cause Notes pollution; D, disease. OE, overexploitation; P, (Source: Committee on Recently Extinct Organisms ( American Museum of Natural History’s 324 Chapter 15 not live in the sea. We are therefore most likely to Introduced species pose a major threat to native identify a species as threatened if it lives in a non- fishes (Courtney and Sauffer 1984; Froese and marine habitat. Nonetheless, McKinney conclud- Torres 1999). They rank as the second most impor- ed that the perception that marine species are less tant threat to freshwater fishes, again mirroring extinction prone is valid. Subsequent work has ar- the threats posed to birds, mammals and plants, gued that the extinction rate of poorly known taxa where introduced species rank as either the third in well-studied regions equals that of the best or fourth most important threat, depending on the studied taxa – mammals (McKinney 1999). Thus, taxon. Over a decade ago, Welcomme (1988) was the appearance of relatively low extinction rates of able to compile data on 1354 introductions of fishes and other marine organisms may simply 237 alien fish species into inland waters. For 72 stem from artefacts of sampling biases and species, it was possible to document the outcome undersampling. of the introduction in terms of impacts on the environment, including native fish species. 15.3.3 Causes of extinction Twenty-five of these introduced species had harm- ful impacts, including reduction or elimina- Human activities have been held responsible for tion of native fish populations, introductions of all known cases of recent extinction (Fig. 15.3). parasites to native fishes, physical disturbance of Remember that, so far, all known extinctions have habitats, or severe depletion of native fishes due occurred in freshwater fishes (Section 15.3.1). to predation. Habitat alteration is the most important problem, Introductions can also cause problems through and includes construction of dams and channels, direct and indirect interactions with native popu- as well as siltation and water extraction (e.g. lations of the same species (Ward, Chapter 9, Vol- Minkley and Deacon, 1991; Miller et al. 1999). The ume 1). Escapes of farmed fishes are widespread; World Conservation Union’s most recent Red List for example 20–30% of spawning Atlantic salmon (Hilton-Taylor 2000) also lists habitat destruction (Salmo salar) in Norway are of farmed origin. Vari- as the most important cause of threats to birds, ous studies have shown genetic differences be- mammals and plants. tween farmed and wild salmon in important

80

70

60

50

40

Percentage 30

20 Fig. 15.3 Percentages of fish extinctions attributed to specific 10 causes. The percentages pertain to a total of 70 species of fish, all 0 occurring in freshwater habitats (see text) but excluding Lake Pollution Victoria cichlids. Many species Overfishing Hybridization were affected by more than one

Habitat alterationIntroduced species Disease/parasites problem. (Source: from Harrison Deliberate eradication and Stiassny 1999.) Conservation and Exploitation 325

fitness traits (e.g. Einum and Fleming 1997; Flem- The problems faced by Lake Victoria cichlids ing and Einum 1997). An ambitious field study in exemplify the typical situation whereby fish popu- Norway showed that farm fish were competitively lations are rarely threatened by a single process. inferior to wild fish, with a lifetime reproductive Instead, the direct impacts of fishing, as well as success of 16% of that of wild fish (Fleming et al. indirect impacts related to fishing, often com- 2000). Males were particularly inferior, courting bine with various forms of habitat degradation to females about one-third as often as did wild males, threaten species (Beverton 1992). These problems and obtaining only 24% of the breeding success of are particularly acute in freshwater bodies and in wild males. Thus, the main route of gene flow was coastal zones, as these waters are the recipients of from wild males mating with farm females. Over- virtually every form of human waste (Moyle and all, farm fish contributed to a reduction in produc- Leidy 1992). In North America, multiple factors tivity by wild fish by over 30%. were implicated in 82% of extinctions of 27 Pressures to farm genetically modified fishes species and 13 subspecies during the past 100 years have brought concerns about introduction of (Miller et al. 1999). alien species and genes into uncharted waters. For Although there is a paucity of known extinc- example, transgenic salmonids have been created tions of marine fish species, we can gain some with the objective of enhancing growth rates of insights into differences between marine and fish used in aquaculture (e.g. Devlin et al. 2001). freshwater environments in processes that may The environmental impacts of escapes to the lead to extinction by making comparisons among wild are unknown, but the lessons from studies of taxa. A comparison of threatened species of fresh- farmed fish described above suggest that concerns water fishes confirms the findings for extinction would be well founded (Hedrick, 2001). Thus, risk, with species most heavily threatened by habi- while no species is known to have become extinct tat loss and introductions (Froese and Torres 1999). as a result of competition with non-native popula- Indeed, there is a correlation between the number tions of the same species, hybridization between of introductions and the number of IUCN threat- wild and domesticated fishes has certainly led to ened species at a national level. In contrast, marine the loss of genetically distinct native populations, introductions have not been linked to any extinc- which we expect to accelerate with the use of tions of native fauna. An analysis of the relative transgenic fishes. There may be impacts on wild importance of different causal factors of extinc- populations through competitive interactions, as tions in the Wadden Sea over the past 2000 years well as genetic introgression, because it is doubtful suggested that overexploitation was responsible that sterility can be achieved with 100% success. for the loss of six species of fishes, with habitat loss Overfishing ranks as the third most important responsible for the loss of four additional species cause of extinction of fishes (Fig. 15.3). The impor- (Wolff 2000). These two factors together are tance of exploitation in general rises if we consider thought to have contributed to the loss of two secondary activities related to it, such as introduc- more fish species, and pollution together with tions of alien species. The most notorious case has habitat loss has caused the loss of one more been the introduction of the Nile perch (Lates species. niloticus) to a fishery in Lake Victoria (Witte et al. 1992). This predatory species has had a devastating impact on rock-dwelling haplochromine cichlids, though estimates of the exact number of species 15.4 EXPLOITATION AS lost are severely hampered by uncertainties about A CAUSE OF DECLINES taxonomic relationships, incomplete sampling, AND EXTINCTION and the role of additional threats such as ecosys- tem alterations due to eutrophication and intro- Comparatively little is known of the disappear- duced macrophytes (Harrison and Stiassny 1999). ances of some fishes. For example, declines of the 326 Chapter 15 endemic New Zealand grayling (Prototroctes as the Georges Bank haddock and various cod oxyrhynchus) were first noted in the year in which stocks on both sides of the Atlantic (Fig. 15.1). The the species was described, 1870. This species is decline of the sablefish (Anoplopoma fimbrae) off presumed to have become extinct by 1930 from a the west coast of the United States provides anoth- combination of habitat degradation due to defor- er example, with stocks reduced by half over a estation plus the impacts of introduced salmonids. period of 27 years (Fig. 15.1f). Exploitation may not have been a major culprit Declines have often been assisted by un- here, though neither of the two proposed explana- favourable environmental conditions, as illustrat- tions is conclusive (McDowall 1996). ed by the short-lived early maturing herring-like fishes – clupeoids. For example, Peruvian anchove- 15.4.1 Targeted species ta stocks crashed dramatically in 1972 as a result of a strong El Niño event that moved warmer waters The common sturgeon (Acipenser sturio) is now nearshore and led to the shutdown of upwelling, captured only rarely in Europe. Historically this consequently reducing availability of their plank- species was highly regarded as a food fish, con- tonic food (Fig. 15.1c) (Glantz 1979). Intense fish- tributing 70% of the fish consumed around the ing continued on the now more vulnerable schools southern Baltic Sea during the 7th–8th century. that had been pushed shorewards. The anchoveta By the 11th century this species was extinct in has taken 25 years to recover, and this process was the Netherlands and by the 12th century sturgeon hindered by smaller El Niño events after the initial were so rare that all captures in the United King- collapse. There are numerous other examples of dom and France were reserved for kings. So by the combinations of exploitation and environmental 14th century sturgeon had been reduced from a problems causing collapses of stocks, such as common to a quantitatively insignificant portion the Monterey Bay sardine fishery (Chapter 1, of the fish catch (Hoffmann 1996). International this volume) and the southern North Sea herring trade in this species has been restricted under the fishery. Convention on International Trade in Endangered A marine fish that is listed under the United Species (CITES) since 1983 and the species is listed States Endangered Species Act is the the totoaba under the IUCN Red List as critically endangered (Totoaba macdonaldi). The plight of this species and it is protected under Appendix III of the Berne highlights several impacts of exploitation and the Convention. All other species of sturgeons and ways in which they interact with other environ- paddlefishes (Acipenseriformes) are threatened by mental problems. The totoaba is a huge species of multiple factors including overexploitation for croaker (Sciaenidae) which is restricted entirely to caviar, habitat loss through dams, channelization, the northern half of the Gulf of California. In the degradation of spawning habitat and pollution. To early 20th century, when directed fishing on the date one species and one population of Acipenseri- species first began, each year the fish migrated in formes are listed as extinct, 6 species are critically huge shoals, following the coast north to their endangered, 10 species are endangered and 7 spawning grounds at the mouth of the Colorado are listed as vulnerable (Hilton-Taylor 2000). Two River. The fish were enormous, reaching lengths species, Acipenser sturio and A. brevirostrum, are of more than 2m and weights upwards of 100kg listed in Appendix I of CITES. At the time of writ- (Cisneros-Mata et al. 1995, 1997). Initially, they ing, Russia, Kazakhstan and Azerbaijan have de- were caught simply for their swim bladders, which clared a complete moratorium on the capture of were sent to the Far East for thickening soups. The sturgeon species from the Caspian Sea, and Iran bodies were piled on the shore to rot or were used exercises strict controls. as fertilizer. It was only in the 1930s that a Direct exploitation has caused the collapse of market developed in California for the meat and many targeted fish stocks. There are some cases the fishery prospered, peaking at 2300 tonnes per where fishing alone has caused the collapse, such year in 1942 (Cisneros-Mata et al. 1995). Conservation and Exploitation 327

Totoaba were just too easy to catch. Sometimes the status of the many species that are rarer than the shoals ran so thick and close to the coast that bocaccio. the fish could be pitchforked out of the water. But Artisanal fisheries are often thought to be less gill-net fisheries began the real decline and the damaging than industrial ones, but there is exten- fishery soon crashed, falling to just 59 tonnes in sive evidence that they also cause species extirpa- 1975. Fishing for totoaba has been banned in tions, which may be the first steps on the road to Mexico since 1975, but the species has continued extinction (Roberts 1997). What is important is not to decline due to bycatch in gill-nets set for other the fishing methods employed, but the intensity of species, and increasing bycatch of juveniles by fishing. Tropical reef fisheries are often pursued at shrimp trawlers near the Colorado River estuary a subsistence level and may be the last resort of (Roberts and Hawkins 1999). Added to the prob- poor people in developing countries (Polunin and lems of bycatch, the estuarine nursery habitat Roberts 1996). The rapid expansion of human pop- has been transformed over this period from ulations in coastal areas has driven enormous in- brackish to predominantly saline, a consequence creases in exploitation rates. In the Caribbean, for of abstraction of water upstream for irrigation. example, several of the larger species of groupers The case of the totoaba demonstrates how marine have been extirpated from intensively fished is- species can be reduced from prolific abundance to lands such as St Lucia and Dominica (Hawkins and the edge of extinction just as surely as terrestrial Roberts, unpublished data; see also Section 15.5.2, animals. below). Such extirpations have also been docu- It is relatively easy to track declines of impor- mented in the Indo-Pacific, especially in southeast tant target fishery species such as the totoaba. It is Asia where the growing live-food fish trade targets much more difficult to document declines in non- larger species like groupers (Bryant et al. 1998). Ex- target species, or species of minor commercial im- tirpations have even been recorded in subsistence portance. For example, the smalltooth sawfish has fisheries, where fish and other organisms are cap- never been of much commercial interest. Like the tured solely for local consumption. Extirpations of totoaba, this estuarine species has been badly hit the giant clam (Tridacna gigas) and the bumphead by bycatch in fish nets and habitat degradation. It parrotfish (Bolbometopon muricatum) have been has been extirpated sequentially from estuaries on documented from subsistence fisheries of the the Atlantic and Gulf of Mexico coasts, and is now isolated Lau Island group, Fiji (Lewis et al. 1988; largely restricted to a few lagoons off the Florida Dulvy and Polunin unpublished data). Human coast (Anon. 2001). populations on these islands have not expanded Another example of severe population declines significantly and they still have poor trade links involves rockfishes (Scorpaenidae) off the west (Zann 1992). coasts of Canada and the United States. This com- So widespread is overexploitation on coral reefs plex of 72 species supports important commercial that even species with large geographic ranges may and recreational fisheries (Yoklavich 1998; Glavin be at risk of extinction. Morris et al. (2000) recently 2001). However, only a few are common enough examined the status of 85 species of epinephelinae for the US National Marine Fisheries Service or the groupers (Serranidae) that inhabit coral reefs. They Canadian Department of Fisheries and Oceans to found evidence for serious declines in 37 species, be able to collect reliable data on their rates of two of which they proposed for ‘endangered’ status capture. Those data show such steep declines in in the World Conservation Union’s Red List and a abundance for several species that extinction is further 35 as ‘threatened’. possible. One of them, the bocaccio (Sebastes pau- Hawkins et al. (2000) drew attention to the fact cispinis), has been added to the World Conserva- that there are far more species of small island en- tion Union’s Red List of Threatened Animals, but demics than had been believed previously. They it is far from being the least common rockfish, or found that nearly 10% of a sample of 1677 coral the most threatened. We simply lack good data on reef fish species had geographic ranges <50000km2 328 Chapter 15

(equivalent to an area of reef habitat of perhaps as other species are threatened by industrial-scale little as a few hundred square kilometres). Many fisheries. restricted-range species are small, like damselfish and wrasses, but direct exploitation for the orna- 15.4.3 Ecosystem impacts mental fish trade may put some of them at risk. For example, the Banggai cardinalfish (Apogon kaud- Exploited species are connected to many others by erni) has an extremely limited distribution in cen- reticulate webs of behavioural and trophic interac- tral Indonesia, where it has been heavily collected tions, the simplest of which are direct interactions for the aquarium trade (Allen 2000). As it can be such as predation, competition and mutualisms bred in aquaria, it is unlikely to go extinct, but it (Chapters 11–16, Volume 1). Furthermore, there could disappear from the wild. are many less direct interactions through which exploitation, pollution and dam construction may 15.4.2 Non-target species have profound and less predictable consequences (Menge 1995; Pinnegar et al. 2000). In freshwater Many other species of vertebrates, fishes and systems eutrophication can lead to phase shifts invertebrates are captured alongside the target from macrophyte- to phytoplankton-dominated species in the relatively non-selective fishing production systems (Strong 1992; Pace et al. 1999; gears commonly used, predominantly trawl nets Scheffer et al. 2001). Such phase shifts can stem on temperate shelves and encirclement nets, traps from trophic cascades, where the removal of a top and hook-and-line gears in reef fisheries. In the predator, which controls the abundance of a herbi- north Australian prawn trawl fishery there are 411 vore, affects the abundance of key basal species, species of fish bycatch alone (Stobutzki et al. 2001). such as algae (Pinnegar et al. 2000; Kaiser and Many species are simply thrown back, while oth- Jennings, Chapter 16, this volume). In hard sub- ers are retained as they have some commercial strate marine ecosystems trophic cascades have value, such as skates. The North American barn- been found to control coral–algal abundance in door skate (Dipturus laevis) (Casey and Myers tropical systems and kelp–coralline algal abun- 1998) and the European common skate (D. batis) dance in temperate systems. One of the key prob- (Rijnsdorp et al. 1996; Walker and Hislop 1998) lems is that there is currently no way of predicting have similar problems due to bycatch in trawl fish- which species have critical ecosystem roles. eries directed at groundfish stocks. Brander (1981) Fishing down food webs, the removal of top conceded that we may have to accept the possible predatory fishes, and subsequent targeting of extinction of such low-value bycatch species as a species further down the food chain, is globally consequence of capturing more valuable species. widespread; although the details can be disputed, It was suggested that the barndoor skate may be the pattern is consistent (Caddy et al. 1998; Pauly close to extinction, although recent surveys sug- et al. 1998a,b). Unfortunately, the ecosystem gest a comeback, restricted to large, no-trawl areas implications of removing top predators (such as on Georges Bank (S. Murawksi, personal commu- sharks) is at best unclear and at worst may be idio- nication). The common skate is restricted to a few syncratic or unpredictable (Stevens et al. 2000). de facto refuges where the bottom is too rough to One suggested consequence of fishing down food fish. Dulvy et al. (2000) argued that the problem of webs is that it could lead to ecosystems dominated skate extirpations is more general than for these by microbial loops (Jackson 2001). There have been species alone, but the problems of detecting de- various pathogen outbreaks resulting in the die- clines and extirpations of individual species have offs of Caribbean seagrass beds, corals and remain- usually been masked by the lumping of different ing herbivores, which may be connected with species into general categories in fishery statistics overexploitation of megafauna, such as manatees, – a common practice for low-value fishes. We are turtles, large predatory fishes and herbivores only just beginning to look into how many (Jackson 1997; Wing 2001). Conservation and Exploitation 329

Recent attention has focused on the distur- especially for species with small ranges. Many of bance effects of trawl gears on the benthos. Heavy, the restricted-range coral reef fish species docu- mobile trawl gears resuspend sediments and kill or mented by Hawkins et al. (2000) had ranges injure benthic invertebrates, which are then eaten overlapping regions where reef habitats are under by scavengers (Hall 1994; Kaiser and Spencer 1996; growing pressure. For example, the damselfish Jennings and Kaiser 1998). This has resulted in de- (Chromis pelloura) is restricted entirely to the far clines of benthic invertebrates (Collie et al. 2000). northern Gulf of Aqaba in the Red Sea, where it is Recently, an analysis of benthic invertebrate com- hemmed in by intensive coastal development. The munity production:biomass ratios along a fishing splendid toadfish (Sanopus splendidus) occurs gradient has demonstrated that trawling has nega- only on reefs around the island of Cozumel in tive impacts on secondary benthic production the Mexican Caribbean, an area undergoing rapid (Jennings et al. 2001a). tourism development. Habitat degradation is responsible for estuarine Recently, it has become clear that coral reef species topping the list of threatened species in the environments are highly sensitive to seawater sea, largely through human development and pol- warming associated with global climate change lution (Roberts and Hawkins 1999). Estuarine and (Reaser et al. 2000). The year 1998 saw the most nearshore habitats like salt-marshes, mangroves widespread coral bleaching and subsequent mor- and seagrass beds provide critical spawning and tality so far documented. Coral mortality reached nursery habitats for a wide range of species, includ- 70–90% throughout large regions of the Indian ing many that we exploit. Such habitats have been Ocean (Wilkinson 2000). Such large-scale habitat cleared extensively for aquaculture, agriculture, degradation, especially if combined with other development and timber. The US has lost more human pressures such as overfishing, could trigger than 50% of its salt-marshes during the last centu- extinctions. Indeed theoretical estimates of the ry (Agardy 1997), while many countries in south- number of extinctions likely to have been caused east Asia have cleared as much as 80% of their by coral bleaching have been calculated using the mangroves in recent decades (Spalding 1998). By species–area curve method. This approach indi- interrupting critical phases of the life cycles of cates that about 1000 species may already have species, habitat destruction has caused many been lost if we accept the most conservative esti- species to decline. Estuaries are also foci for intro- mate of a loss of 5% of the world’s reefs (Carlton et ductions of invasive species that may threaten na- al. 1999). The key caveat is that bleaching causes tive fauna (Cohen and Carlton 1998). For example, hard and soft coral loss alone, rather than complete the spotted handfish (Brachionichthys hirsutus) is loss of all reef microhabitats. There are bound to be restricted entirely to the Derwent River estuary in species thriving on reefs with little or no remain- Tasmania and may be threatened by predation on ing coral as a consequence of bleaching. its benthic egg clusters by starfish (Roberts and Hawkins 1999). Other marine habitats are also being damaged and destroyed at accelerating rates. For example, 15.5 WHAT RENDERS the widespread use of highly destructive blast fish- SPECIES SUSCEPTIBLE ing in southeast Asia is converting diverse and TO OVERFISHING? complex habitats to rubble (Cesar et al. 1997). Local impacts and habitat loss such as the con- From the foregoing discussion and various reviews struction of a military airbase in Castle Harbour, (e.g. Beverton 1992; Roberts and Hawkins 1999; Bermuda, has led to the local extirpation of 10–14 Reynolds et al. 2001b), we have picked out five key species of reef fishes (Smith-Vaniz et al. 1999). Fur- features of the biology of fishes and the motiva- thermore, habitat destruction is sufficiently wide- tions of fishers that render fish populations suscep- spread that it could cause global extinctions, tible to overfishing. 330 Chapter 15

15.5.1 Catchability remains high havioural resilience in the population as a whole. as population size decreases There is growing evidence of assortative mating and genetic substructuring on even spatially Species that form shoals can still be targeted prof- diffuse breeding grounds of more wide-ranging itably by fishers even as the total population sizes species such as cod (Hutchings et al. 2000; Nordei- decline (Pitcher 1995; Mackinson et al. 1997). de and Folstad 2000; Ruzzante et al. 2000). Such Thus, species such as herring and Peruvian an- aggregations are easily targeted. choveta can still be caught effectively by purse The classic example of vulnerability due to fish- seines as populations decline. This continuing effi- ers targeting spawning aggregations concerns the ciency works against the old concept that as fish Nassau grouper (Epinephelus striatus) (Sadovy and became rare, they would become unprofitable and Eklund 1999). In some areas of the Caribbean ap- therefore subject to lower mortality (see also next proximately 90% of commercial and recreational section). landings of this species came from spawning sites. Species that migrate through physical bottle- Aggregations no longer form at some previously necks, such as diadromous salmonids, shads and known traditional sites in Belize, the Dominican sturgeons, are susceptible to high mortality due to Republic, Honduras, Mexico and the Bahamas, focused fishing effort and pollution outputs associ- while none of the previously known aggregations ated with the dense human populations of river remain in Bermuda, Puerto Rico, or the US Virgin mouths (McDowall 1992). The impact of dams on Islands. In Cuba, 21 aggregations were known in freshwater fish populations, particularly the At- the 1800s. Today, however, only one of these sites lantic salmon (Salmo salar), has been known since is known to remain in use. The status of many before 1214, when a Scottish statute required that other spawning aggregations is unknown. all dams be fitted with an opening (Hoffmann 1996; unpublished data). Damming and the intro- 15.5.2 Fish are highly valuable duction of dikes also contributed to the decline of the common sturgeon. This was confirmed by the Some fish derive their value not only as a source of revival of catches after floods destroyed barriers in protein, but also from cultural or social values the southern Baltic in the 1400s.The catches re- leading to premium prices. Buyers capable of pay- mained high until new reclamation works in the ing for rare species may advertise their wealth and 1800s (Hoffmann 1996; unpublished data). Con- social status. The reservation of caviar and stur- servation of salmonids remains a serious concern, geon for English and French nobility in the 14th as their discrete freshwater spawning populations century is testament to the deep-rooted existence remain subject to a variety of threats, including of this behaviour, which continues today in fine forestry and damming (Jonsson et al., 1999). To restaurants. Raw tuna or sashimi is highly coveted date, 55 evolutionarily significant units (ESUs) are by rich restaurant-goers in Japan. The most prized recognized for the seven anadromous species of species is the southern bluefin tuna. It set a new Pacific salmon, of which 23 ESUs are listed as price record early in 2001, when US$178000 was endangered under the US Endangered Species paid for one individual (Watts 2001). Western At- Act (R.S. Waples, personal communication). lantic bluefin tuna (Thunnus thynnus) has sold for Species with limited physiogeographic ranges, up to US$83500 for a single individual. It provided living in small catchments, with specific breeding 2400 servings of sushi, worth US$180000. Anoth- and feeding habitats appear to be particularly vul- er expensive fish is the giant yellow croaker (Baha- nerable to extinction in freshwater habitats ba taipingensis), which has been exploited for its (Angermeier 1995; Parent and Schriml 1995). In swimbladder in the South and East China seas the sea critical habitats such as locations where from Shanghai to Hong Kong. In recent years the groupers aggregate to spawn are vulnerable. The swimbladder, or maw, has been called ‘soft gold’ loss of subpopulations may result in the loss of be- due to market prices of US$20000–64000 per kg Conservation and Exploitation 331

(Sadovy and Leung 2001). It is highly valued for its without any regulations, and these species rarely medicinal properties, and as a health tonic, and is appear in catch statistics. Non-target species thus typically boiled and drunk as a soup. Despite the inhabit the poorly known underworld of fisheries near extinction of this species, 100–200 boats still conservation. They are either discarded at sea or, target its historical spawning sites in the hope of if landed, they often fetch a lower price than the netting a windfall (Sadovy and Leung 2001). species being targeted. Therefore they do not at- Unfortunately for conservation efforts the high tract much attention from assessment biologists prestige and price of rare species often widen the or managers. geographical net to meet demand, thereby threat- ening even those species that have large ranges. 15.5.4 Life histories result in The humphead, maori or Napoleon wrasse (Cheili- low productivity nus undulatus) is a highly valued commodity in the live reef fish trade centred in Hong Kong Species with long generation times, low natural and reaching Taiwan, Singapore and China. It is mortality rates and slow body growth are expected worth a retail price of up to US$130 per kg, with to be less able to withstand elevated mortality (re- parts of this fish reaching much higher prices (Lau viewed by Musick 1999b; Reynolds et al. 2001b). and Parry-Jones 1999). As southeast Asian sources This has been shown on theoretical grounds (e.g. have been overfished, demand has led to fishing ex- Adams 1980; Kirkwood et al. 1994; Pope et al. plorations and export operations as far away as 2000), and it has been borne out by comparisons the Seychelles, Fiji and Kiribati (Bentley and among diverse taxa (e.g. Jennings et al. 1998, Aumeeruddy 1999). This increasing price associat- 1999a, 1999b; Dulvy et al. 2000). Thus, sharks and ed with biological and market rarity has meant rays feature prominently in the Red List of Threat- that it is economically viable to fly fish to markets ened Animals (Hilton-Taylor 2000), on the basis from such distant sources. This species is now list- of severe population declines under exploitation. ed as threatened under IUCN criteria (Donaldson Other examples of species with life histories that and Sadovy, 2001). Given the role that increas- are incompatible with elevated mortality include ing ‘rarity value’ has had in depleting terrestrial sturgeons, rockfishes (Sebastes) (Section 15.4.1) species such as African elephants, rhinoceroses, and orange roughy (Hoplostethus atlanticus) in tigers and musk deer, it is difficult to see how cur- New Zealand, which reaches maturity in its twen- rent aquatic conservation methods can protect ties to thirties and may live to a maximum age of such species. International trade can be prevented 150 years (Smith et al. 1995; Horn et al. 1998). The through CITES Appendix I listing, if enforcement coelacanth (Latimeria chalumnae) also fits this is adequate. However, this would not prevent trade category. While most details of its life history re- within countries, and this measure would only main unknown, this species has the lowest meta- be introduced after species had already declined bolic rate known for any fish (Fricke and Hissmann seriously. 2000). This suggests a very slow life history, which would render the species susceptible to the mortal- 15.5.3 Fish are susceptible to ity that individuals suffer as a result of bycatches capture as non-target species in deepwater artisanal fisheries in the Comoros and in Indonesia (reviewed by Fricke 2001). If fish are caught as a byproduct of other activities, they again defy the hope that unprofitability at low 15.5.5 Per capita recruitment population sizes might protect their populations. decreases as population size decreases We have already described several examples of this problem for species such as skates and rays (Sec- Depensation, called the ‘Allee effect’ in terrestrial tion 15.4.2). Stevens et al. (2000) noted that rough- systems, occurs when there is a positive relation- ly 50% of elasmobranchs are taken as bycatch ship between individual productivity and popula- 332 Chapter 15 tion size (Myers, Chapter 6, Volume 1). In fisheries, introductions of alien species. These problems are depensation can occur due to a reduced ability particularly acute in freshwater habitats. Further- to aggregate and find mates, reduced fertilization more, many conservationists point to the repeated success, or increased predation rates (reviewed by failure of fisheries management to maintain ade- Petersen and Levitan 2001). The concern here is quate populations, even when fisheries are under that as fish stocks are pushed downwards, they the ‘control’ of single nations or unions of member may fall over a cliff of recruitment from which states. The northern cod stock(s) off Newfound- they cannot climb back. A Bayesian analysis of vari- land, which has still not shown convincing signs of ous fish stocks by Liermann and Hilborn (1997) recovery since a ban on fishing was imposed in showed that scatter in relationships between stock 1992, is the most recent notorious example. Some sizes and recruitment cause considerable difficul- resource managers counter that there are still ties in detecting depensation, if it exists. For a millions of Atlantic cod in the sea, and that you number of species of cod relatives (Gadiformes), couldn’t kill off this species if you tried. At this flatfishes (Pleuronectiformes) and herring rela- point, conservationists tend to bring up the fate of tives (Clupeiformes), the authors found that the the passenger pigeon (Ectopistes migratorius), tails of probability distributions for the likelihood which once numbered between 3 and 5 billion in of depensation extended well into the depensation the mid-1800s, but became extinct in 1914. The range. However, attempts to quantify the preva- debate continues from there. lence of depensation based on current stock– How did we come to this? We blame cod and recruitment (S–R) models may be too conserva- haddock. Specifically, it was their listing as ‘vulner- tive. This is because current models assume that able’ (to extinction) by the Red List of Threatened recruitment is zero only when the stock is extinct. Species (IUCN 1996) that triggered a confrontation However, Allee effects, by definition, involve low between some conservationists and some resource recruitment despite the continued presence of biologists. The species were listed on the basis spawners. Therefore, future assessments of the ex- of their population declines within the previous istence of depensation in fishes may need a new three generations. Rate of decline is one of five cri- family of S–R models, which do not necessarily teria under which a species can be listed as threat- have their origin at zero (Frank and Brickman ened. The others are small distribution combined 2000). with declines or fluctuations, small populations combined with declines, very small or restricted ranges, and quantitative population models that 15.6 CONSERVATION MEETS yield pessimistic outcomes. A caveat was pub- SUSTAINABLE USE lished with the Red List, which noted that its des- ignations may not be appropriate for fish species Conservationists traditionally worry about pre- that are subject to management by fisheries. How- venting extinction, whereas resource managers in ever, this did not save the Red List’s rate-of-decline fisheries and forestry traditionally ignore extinc- criterion from criticism by many fisheries biolo- tion risk, and worry instead about obtaining high gists (e.g. Musick 1999a; Butterworth 2000). sustained yields (Mace and Hudson 1999). Of One criticism of the Red List’s treatment of course, if the yield objective is achieved, we won’t exploited fishes is that traditional management have to worry about extinction! However, most of practice suggests that populations should decline the world’s exploited fish species are not being as- to about 50% of virgin population sizes in order sessed or managed. Therefore, extinction cannot to maximize productivity (e.g., Schnute and be ruled out for many species that have the vulner- Richards, Chapter 6, this volume). So in theory, a able characteristics listed in the previous section, perfectly well-managed abundant species could be especially if exploitation exacerbates other prob- brought quickly to 50% of its population size and lems such as habitat degradation, pollution and provide a maximum yield, only to be listed by a Conservation and Exploitation 333 globally recognized conservation body as threat- decline required for a listing under this proposal ened with extinction. In practice, of course, one are not sufficiently precautionary (Reynolds et al. could suggest that this is all a moot point, because 2001b). For example, a species deemed to have high there are few examples of any virgin population of productivity would be allowed to decline by 99% fishes being brought down to 50% (and not beyond) over 10 years or three generations, whichever is in the controlled way envisaged by this hypotheti- longest, before it would be listed as ‘vulnerable’ cal scenario. But even if this situation is only hypo- (Musick 1999a). thetical, it does suggest an anomaly, which the In the meantime, after a series of scientific IUCN has been trying to fix (see below). workshops, the World Conservation Union Another problem that has been raised is that it brought out a refined set of criteria in 2001. Among is difficult to distinguish between ‘real’ declines various refinements there were two key changes. and population fluctuations in many marine fishes First, for any taxon to be listed as ‘vulnerable’ (the (Butterworth 2000). A recent study by Hutchings lowest threat status), the species must have de- (2001b) suggested that fish populations do not fluc- clined by 30% (formerly 20%) over the previous 10 tuate more than populations of terrestrial animals. years or three generations (whichever is longest). So, while the problem certainly deserves to be Second, and particularly relevant to exploited taken seriously, it is not unique to fishes. species, higher declines are allowed before trigger- Finally, the high fecundity of many exploited ing a listing if the causes of the decline are under- fish species has been taken to imply that such stood, clearly reversible, and have ceased (IUCN stocks have high potential to bounce back from 2001). Now, the species in our hypothetically con- low numbers (Musick 1999a). The theory and trolled fishery can decline by at least 50% before evidence in support of this assumption have being listed as ‘vulnerable’ (formerly 20%), 70% been questioned (Sadovy 2001). Indeed, Hutchings for ‘endangered’ status (formerly 50%), and 90% (2000) found little evidence of recovery for most of for ‘critically endangered’ status (formerly 80%). the 90 stocks that he has examined. Although his The Red List has taken on the ambitious task of analyses did not take fishing mortality following combining versatility with practicality so that the population declines into account explicitly, subse- same criteria can be applied to a wide variety of quent analyses have shown that recoveries remain plant and animal taxa. Thus, the aim is to draw at- slower than generally expected even when the data tention to species when they show at least one of are restricted to stocks in which fishing mortal- the symptoms of vulnerability towards extinction, ity during the recovery phase is extremely low including severe population declines (Mace and (Hutchings 2001b). Hudson 1999). Defenders of this approach argue After the uproar over the listing of commercial- that the Red List is intended as precautionary flag, ly exploited fishes, there was a careful rethinking not a prescription for action, though it is hoped of the Red List’s main criteria for listing species. that it might lead to careful assessment, and man- CITES has also been grappling with these issues, agement if required. with advice from the FAO (Butterworth 2000), as Perhaps the best way to bridge this divide be- have the Committee on the Status of Endangered tween conservationists and resource managers is Wildlife in Canada (COSEWIC), the Australian So- to incorporate extinction risk explicitly into popu- ciety for Fish Biology and the American Fisheries lation models, so that traditional fisheries refer- Society (Musick 1999a). The American Fisheries ence points such as maximum sustainable yield Society has adopted a two-step listing process, (MSY) can be compared with the probability of ex- under which species with life history characteris- tinction. Several researchers have begun to build tics that imply high productivity and resilience this bridge. Matsuda et al. (1998) used three meth- would be allowed to decline more steeply than the ods to evaluate the probability that the southern IUCN suggests before triggering a listing. Else- bluefin tuna would become extinct (less than 500 where, we have argued that the high rates of individuals) within the next 100 years: a simple 334 Chapter 15 simulation model, a model that used a diffusion exploitation, whereby only a fraction of the esti- approximation of births and deaths, and a fluctuat- mated surplus above the threshold is taken, rather ing age-structured model. None of these found than aiming for maximum exploitation of the sur- support for a high probability of extinction under plus. These models are noteworthy not only for current circumstances, contrary to the IUCN’s their contributions to the theory of exploitation of 1996 and 2000 listing of this species as critically fluctuating populations, but also because they ex- endangered. This was largely due to the large cur- plicitly incorporate extinction risk, unlike tradi- rent population sizes of these fish, despite their tional models such as yield-per-recruit, which are high rates of decline. Matsuda et al. (1998) con- in widespread use (see Shepherd and Pope, Chapter cluded that the decline criterion should be linked 8, this volume, and Sparre and Hart, Chapter 13, to population sizes to give a better reflection of this volume). extinction probability. A direct comparison of traditional fisheries ref- erence points with extinction risk has been made 15.7 WHAT IS NEEDED by Punt (2000). He used a deterministic population TO SAFEGUARD FISH model that compared the level of fishing mortality BIODIVERSITY? that achieves maximum sustainable yield (FMSY) with that which causes the population to become It should be clear from this review that freshwater extinct, defined as 1/1000th of initial popula- and marine biodiversity is seriously threatened. tion size (Fcrash). At this point, depensation might The bright side is that although many populations occur, whereby per capita recruitment might have undergone steep declines no species of decline, thus potentially spiralling the population marine fishes is known for certain to have become downwards to extinction (Section 15.5.5; see also extinct. As we have argued in Section 15.3, there is Myers, Chapter 6, Volume 1). The analyses con- good reason for thinking that we are less able to firmed the expectation that Fcrash was highest for detect extinctions in marine environments. But at highly productive populations, and that depensa- least we can say that no fisheries manager goes to tion greatly reduced the ratio of Fcrash :FMSY. In tests bed at night with a conscience laden with the guilt with case studies, Punt found that in two shark of having seen a species of targeted marine fish species Fcrash was only about twice FMSY. This is become extinct during their watch. worrying because FMSY cannot be estimated with The situation for freshwater fishes is far more much precision in many fisheries, so one cannot be dire in many parts of the world. Direct and indirect certain how far above or below it the actual fishing effects of exploitation, habitat destruction and mortality is. degradation loom large among the drivers of de- A third approach towards bridging conservation cline. As demand for freshwater increases, and and sustainable use has examined the effects of technologies for catching fish have improved, so various exploitation strategies on yields and long- the areas free from exploitation and habitat loss term risks of population collapse and extinction in have diminished. We treat inland water bodies as populations that undergo strong natural fluctua- sources of irrigation and hydroelectric power, at tions (Lande et al. 1995, 1997; reviewed by Lande et the expense of aquatic biodiversity. al. 2001). These studies suggest that if the goal is to maximize the cumulative yield before extinction 15.7.1 Marine reserves or population collapse, the best approach is to use threshold exploitation, whereby populations are Marine reserves are often championed as one fished only when they overshoot their carrying mechanism for protecting marine biodiversity capacities. A more prudent strategy, appropriate (Watling and Norse 1998; NRC 2000; Roberts and when there is considerable uncertainty about Hawkins 2000). The evidence for and against this population sizes, is to use proportional threshold viewpoint is reviewed critically by Polunin Conservation and Exploitation 335

(Chapter 14, this volume). Syntheses of research using lack of information as an excuse for inaction, from around the world have shown that the cre- and using reference points. Reference points ation of reserves closed to fishing does yield rapid include benchmark population sizes or mortality increases in abundance, body size and diversity of rates that are not to be exceeded (‘limit reference marine communities (Mosquera et al. 2000; Côté points’) or which are desirable (‘target reference et al. 2001; Halpern, in press). Reserves can there- points’) (see also Shepherd and Pope, Chapter 8, fore provide an important fishery management this volume). However, for precautionary refer- tool by putting back the refuges that fishing has ence points to be successful it is critical that once a eroded away during the last century. Furthermore, benchmark is agreed upon the conservation goal- fully protected reserves can begin the process of post is not shifted. Powles et al. (2000) outlined habitat recovery from fishing disturbances such as the continuum and overlap of both fisheries man- trawling or blast fishing. agement and conservation benchmarks. As some Because of their promising role in fishery species have declined management has shifted management, reserves are viewed as a potential from one benchmark to the next, allowing popula- ecological–economic win–win tool (Pezzey et al. tions to go from growth overfishing to recruitment 2000; Rodwell and Roberts 2000), providing an overfishing, and up the scale to a critically endan- economic rationale for doing what is also sensible gered listing under IUCN criteria. from a conservation perspective. Current under- standing of marine reserves suggests that they 15.7.3 Targeted management at will deliver maximum benefits when they cover key points in the life history between 20% and 50% of every habitat and biogeo- graphic region of the oceans (NRC 2000). Theoreti- Identifying key points in the life history of organ- cal studies suggest that they will be most effective isms could prove to be a fruitful approach for focus- when established in dense networks consisting of ing often-limited management efforts. However, areas of a few to a few tens of kilometres across, identification of critically important ages, stages, and which are separated by a few to a few tens of habitats or even sexes is still in its infancy. kilometres (Roberts et al., in press). Guesses can be made based on experience; salmon Of course marine reserves will not be sufficient are clearly more vulnerable in estuaries than in the to protect high seas and migratory fish stocks like open sea, and females usually limit populations swordfish, marlin and tuna, and their use in fresh- more strongly than do males. However, we still water has barely been explored. Furthermore, their have a lot to learn. For example, at present man- implementation needs to be augmented by overall agers are unsure as to whether is it better to protect reductions in fishing effort and other technical juvenile skates or mature females. While control measures (Murawski et al. 2000; Wabnitz and of mortality in either life stage would be difficult, Polunin 2001; Polunin, Chapter 14, this volume). one can imagine that closed areas could be applied Otherwise, reserves may simply cause a redistri- to nursery areas or locations that are primarily bution of fishing effort, without leading to reduced used by adults. mortality. One way forward is the application of elasticity analyses, based on demographic matrix models 15.7.2 The precautionary principle (e.g. Kokko et al. 2001). These can be used to deter- and reference points mine the relative importance of various stages of the life cycle for the population growth rate. One of The precautionary principle is now being imple- the first uses of this method was to examine the mented in fisheries management in many parts of conservation method of ‘head starting’, which is the world (FAO 1995). The elements of this princi- widely applied to a bycatch turtle species. This in- ple are simple, such as taking account of uncer- volves enhancing the hatching and survival rate tainty, being cautious with new fisheries, not of turtles, which was previously assumed to be the 336 Chapter 15 critical stage for population growth. However, question about the vulnerability of freshwater demographic modelling and elasticity analyses fishes to extinction, and the potential for exploita- demonstrated that this life stage contributed little tion to exacerbate the threats. There have been to population growth rate compared to the sub- positive developments in the field of fisheries con- adult and adult stages (Heppell et al. 1996). Similarly, servation, including studies of marine reserves, while fishery scientists often focus on understand- ecosystem-based analyses, adoption of precaution- ing what determines survival in the first year of ary reference points that embrace the uncertainty life, elasticity analyses hint that it is more critical that pervades fisheries, and theoretical studies to protect fishes between the first year and matu- that bridge between management targets and risks ration (Heppell et al. 1996). For example, in the of extinction. We look forward to further progress North Sea haddock survival in the first year of life in all of these fields, which should lead to more contributes only approximately 30% to the overall prudent use of fishes as resources while protecting population growth rate, whereas survival from the them and the environment. first year to maturity contributes approximately 60%. We should therefore focus conservation effort on allowing juveniles to reach matur- ACKNOWLEDGEMENTS ity rather than focusing on first-year juveniles (Heppell et al. 1999). Though such approaches are We thank Yvonne Sadovy and Richard Hoffmann data-intensive, the information should be gener- for providing unpublished manuscripts and com- ally applicable to species that have similar life ments. This chapter is dedicated to the memory of histories to those studied so far. Don McAllister, a committed environmentalist There is a critical need for more precautionary and an inspiration to all those who knew him. and ecosystem-based approaches in fishery man- NKD was supported by the UK’s Natural Environ- agement (NRC 1999). What has become clear is ment Research Council. that present approaches to fishery management are too risk-prone, failing to take adequate account of irreducible uncertainties in fishing mortality REFERENCES rates or future environmental conditions (Ludwig et al. 1993; Mangel 2000; Roberts 2000). In combi- Adams, P.B. (1980) Life history patterns in marine fishes nation, reduced fishing effort, extensive use of re- and their consequences for management. Fishery Bulletin 78, 1–12. serves in both marine and freshwater habitats and Agardy, T.S. (1997) Marine protected areas and ocean management focused on critical life stages will do conservation. Texas: Academic Press. much to secure the future of fish species. Allen, G.R. (2000) Threatened fishes of the world: Pterapogon kauderni Koumans, 1933 (Apogonidae). Environmental Biology of Fishes 57, 142. 15.8 CONCLUSIONS Angermeier, P.L. (1995) Ecological attributes of extinction-prone species: loss of freshwater fishes of Virginia. Conservation Biology 9, 143–58. For too long, fisheries biologists and conservation Anon. (2001) At sea, at risk. Science 292, 617. biologists have been attending different confer- Bentley, N. and Aumeeruddy, R. (1999) The live reef ences, publishing in different journals, and worry- fishery in the Seychelles. South Pacific Commission ing about different things. Extinction risk has not Live Reef Fish Bulletin 6, 5–7. usually been a concern in marine fisheries, but the Beverton, R.J.H. (1992) Fish resources; threats and collapses of fish stocks, extirpations and damage to protection. Netherlands Journal of Zoology 42, 139–75. Brander, K. (1981) Disappearance of common skate Raia ecosystems, combined with considerable uncer- batis from Irish Sea. Nature 290, 48–9. tainty about risks of cryptic extinctions, are caus- Bryant, D., Burke, L., McManus, J. and Spalding, M. ing conservationists to ask resource biologists (1998) Reefs at Risk: a Map Based Indicator of Poten- awkward questions. Furthermore, there is no tial Threats to the World’s Coral Reefs. World Conservation and Exploitation 337

Resources Institute, Washington DC, International Dulvy, N.K., Metcalfe, J.D., Glanville, J., Pawson, M.K. Center for Living Aquatic Resource Management, and Reynolds, J.D. (2000) Fishery stability, local Manila, and World Conservation Monitoring Centre, extinctions, and shifts in community structure in Cambridge. skates. Conservation Biology 14, 283–93. Bucher, E.H. (1992) The causes of extinction of the Dulvy, N.K. and Reynolds, J.D. (2002) Predicting extinc- passenger pigeon. Current Ornithology 9, 1–36. tion vulnerability in skates. Conservation Biology 16, Butterworth, D.S. (2000) Possible interpretation 440–50. problems for the current CITES listing criteria in the Einum, S. and Fleming, I.A. (1997) Genetic divergence context of marine species under commercial harvest. and interactions in the wild among native, farmed and Population Ecology 42, 29–35. hybrid Atlantic salmon. Journal of Fish Biology 50, Caddy, J.F., Csirke, J., Garcia, S.M. and Grainger, R.J.R. 634–51. (1998) How pervasive is ‘fishing down marine food FAO (1995) Precautionary Approach to Fisheries. Part 1: webs’? Science 282, 1383. Guidelines on the Precautionary Approach to Capture Carlton, J.T., Geller, J.B., Reaka-Kudla, M.L. and Norse, Fisheries and Species Introductions. FAO Fisheries E.A. (1999) Historical extinctions in the sea. Annual Technical Paper, No. 350, 1. Rome: FAO. Reviews of Ecology and Systematics 30, 515–38. FAO (1999) The State of World Fisheries and Aquacul- Casey, J.M. and Myers, R.A. (1998) Near extinction of a ture 1998. Rome: FAO. large, widely distributed fish. Science 281, 690–2. Fleming, I.A. and Einum, S. (1997) Experimental tests of Cesar, H., Lundin, C.G., Bettencourt, S. and Dixon, J. genetic divergence of farmed from wild Atlantic (1997) Indonesian coral reefs – an economic analysis of salmon due to domestication. ICES Journal of Marine a precious but threatened resource. Ambio 26, 345–50. Science 54, 1051–63. Cisneros-Mata, M.A., Montemayor-Lopez, G. and Fleming, I.A., Hindar, K., Mjølnerød, I.B., Jonsson, B., Roman-Rodriguez, M.J. (1995) Life-history and conser- Balstad, T. and Lamberg, A. (2000) Lifetime success vation of Totoaba macdonaldi. Conservation Biol- and interactions of farm salmon invading a native ogy 9, 806–14. population. Proceedings of the Royal Society of Cisneros-Mata, M.A., Botsford, L. and Quinn, J.F. (1997) London B 267, 1517–23. Projecting viability of Totoaba macdonaldi, a popula- Frank, K.T. and Brickman, D. (2000) Allee effects and tion with unknown age-dependent variability. Ecolog- compensatory population dynamics within a stock ical Applications 7, 968–80. complex. Canadian Journal of Fisheries and Aquatic Cohen, A.N. and Carlton, J.T. (1998) Accelerating inva- Sciences 57, 513–17. sion rate in a highly invaded estuary. Science 279, Fricke, H. (2001) Coelacanths: a human responsibility. 555–8. Journal of Fish Biology 59 (Supplement A), 332–8. Coleman, F.C., Koenig, C.C. and Collins, L.A. (1996) Re- Fricke, H. and Hissmann, K. (2000) Feeding ecology and productive styles of shallow-water groupers (Pisces: evolutionary survival of the living coelacanth Latime- Serranidae) in the Eastern Gulf of Mexico and the con- ria chalumnae. Marine Biology 136, 379–86. sequences of fishing spawning aggregations. Environ- Froese, R. and Torres, A. (1999) Fishes under threat: an mental Biology of Fishes 47, 129–41. analysis of the fishes in the 1996 IUCN Red List. In: Collie, J.S., Hall, S.J., Kaiser, M.J. and Poiner, I.R. (2000) A R.S.V. Pullin, D.M. Bartley and J. Kooiman (eds) To- quantitative analysis of fishing impacts on shelf-sea wards Policies for the Conservation and Sustainable benthos. Journal of Animal Ecology 69, 785–98. Use of Aquatic Genetic Resources. Manila: Interna- Côté, I.M., Mosqueira, I. and Reynolds, J.D. (2001) Effects tional Centre for Living Aquatic Resource Manage- of marine reserve characteristics on the protection ment, pp. 131–44. of fish populations: a meta-analysis. Journal of Fish Glantz, M.H. (1979) Science, politics and economics Biology 59 (Supplement A), 178–89. of the Peruvian anchoveta fishery. Marine Policy 3, Courtney, W.R., Jr. and Sauffer, J.R., Jr. (eds) (1984) Distri- 201–10. bution, Biology, and Management of Exotic Fishes. Glavin, T. (2001) The Conservation of Marine Biological Baltimore, Md.: Johns Hopkins University Press. Diversity and Species Abundance on Canada’s West Devlin, R.H., Biagi, C.A., Yesaki, T.Y., Smailus, D.E. and Coast: Institutional Impediments. Report for the Byatt, J.C. (2001) Growth of domesticated transgenic Sierra Club of Canada. http://bc.sierraclub.ca/ fish. Nature 409, 781–2. Campaigns/Marine/Groundfish_2.pdf. Donaldson, T.J. and Sadovy, Y. (2001) Threatened Glynn, P.W. (1988) El Niño-Southern Oscillation 1982– fishes of the world: Cheilinus undulatus Rüppell, 83: nearshore population, community and ecosystem 1835 (Labridae). Environmental Biology of Fishes 62, responses. Annual Reviews of Ecology and Systemat- 428. ics 19, 309–45. 338 Chapter 15

Hall, S.J. (1994) Physical disturbance and marine benthic Hutchings, J.A. and Myers, R.A.M. (1994) What can be communities: life in unconsolidated sediments. learned from the collapse of a renewable resource? Oceanography and Marine Biology 32, 179–239. Atlantic cod, Gadus morhua, of Newfoundland and Halpern, B. In press. The impact of marine reserves; do Labrador. Canadian Journal of Fisheries and Aquatic reserves work and does reserve size matter? Ecological Sciences 51, 1216–46. Applications. ICES (2001) Report of the ICES Advisory Committee on Harrison, I.J. and Stiassny, M.L.J. (1999) The quiet crisis: Fishery Management, 2000. Co-operative Research a preliminary listing of the freshwater fishes of the Report No 242. (see also http://www.ices.dk/globec/). world that are extinct or ‘missing in action’. In: R.D.E. Irish, K.E. and Norse, E.A. (1996) Scant emphasis on MacPhee (ed.) Extinctions in Near Time. New York: marine biodiversity. Conservation Biology 10, 680. Kluwer Academic/Plenum, pp. 271–331. IUCN (1996) 1996 IUCN Red List of Threatened Ani- Hawkins, J.P., Roberts, C.M. and Clark, V. (2000) The mals. Gland, Switzerland: International Union for the threatened status of restricted range coral reef fish Conservation of Nature. species. Animal Conservation 3, 81–8. IUCN (2001) Revised Criteria for Threatened Species. Hedrick, P.W. (2001) Invasion of transgenes from salmon Gland, Switzerland: International Union for the or other genetically modified organisms into the Conservation of Nature. natural environment. Canadian Journal of Fisheries Jackson, J.B.C. (1997) Reefs since Columbus. Coral Reefs and Aquatic Sciences 58, 841–4. 16, S23–S32. Heppell, S.S., Crowder, L.B. and Menzel, T.R. (1999) Life Jackson, J.C. (2001) What was natural in the coastal table analysis of long-lived marine species with oceans? Proceedings of the National Academy of Sci- implications for conservation and management. In: ences 98, 5411–18. J.A. Musick (ed.) Life in the Slow Lane: Ecology Jennings, S., Dinmore, T.A., Duplisea, D.E., Warr, K.J. and Conservation of Long-lived Marine Animals. and Lancaster, J.E. (2001a) Trawling disturbance can Bethesda, Md: American Fisheries Society Symposium modify benthic production processes. Journal of Ap- 23 pp. 137–47. plied Ecology 70, 459–75. Heppell, S.S., Limpus, C.J., Crouse, D.T., Frazer, N.B. and Jennings, S., Greenstreet, S.P.R. and Reynolds, J.D. Crowder, L.B. (1996) Population model analysis for the (1999a) Structural change in an exploited fish commu- loggerhead sea turtle, Caretta caretta, in Queensland. nity: a consequence of differential fishing effects on Wildlife Research, 23, 143–59. species with contrasting life histories. Journal of Ani- Hilton-Taylor, C. (2000) 2000 IUCN Red List of Threat- mal Ecology 68, 617–27. ened Species. Gland, Switzerland: International Union Jennings, S. and Kaiser, M.J. (1998) The effects of fishing for the Conservation of Nature. on marine ecosystems. Advances in Marine Biology Hoffmann, R.C. (1996) Economic development and 34, 201–352. aquatic ecosystems in medieval Europe. American Jennings, S., Kaiser, M.J. and Reynolds, J.D. (2001b) Ma- Historical Review 101, 631–69. rine Fisheries Ecology. Oxford: Blackwell. Horn, P.L., Tracey, D.M. and Clark, M.R. (1998) Between- Jennings, S., Reynolds, J.D. and Mills, S.C. (1998) Life his- area differences in age and length at first maturity of tory correlates of responses to fisheries exploitation. the orange roughy Hoplostethus atlanticus. Marine Proceedings of the Royal Society of London B 265, Biology 132, 187–94. 333–9. Hutchings, J.A. (2000) Collapse and recovery of marine Jennings, S., Reynolds, J.D. and Polunin, N.V.C. (1999b) fishes. Nature 406, 882–5. Predicting the vulnerability of tropical reef fishes to Hutchings, J.A. (2001a) Conservation biology of marine exploitation with phylogenies and life histories. Con- fishes: perceptions and caveats regarding assignment servation Biology 13, 1466–75. of extinction risk. Canadian Journal of Fisheries and Jonsson, B., Waples, R.S. and Friedland, K.D. (1999) Aquatic Sciences 58, 108–21. Extinction considerations for diadromous fishes. ICES Hutchings, J.A. (2001b) Influence of population decline, Journal of Marine Science 56, 405–9. fishing, and spawner variability on the recovery of Kaiser, M.J. and Spencer, B.E. (1996) The effects of beam- marine fishes. Journal of Fish Biology 59 (Supplement trawl disturbance on infaunal communities in differ- A), 306–22. ent habitats. Journal of Animal Ecology 65, 348–58. Hutchings, J.A., Bishop, T.D. and McGregor-Shaw, C.R. Kent, G. (1998) Fisheries, food security, and the poor. (2000) Spawning behaviour of Atlantic cod, Gadus Food Policy 22, 393–404. morhua: evidence of mate competition and mate Kirkwood, G.P., Beddington, J.R. and Rossouw, J.A. choice in a broadcast spawner. Canadian Journal of (1994) Harvesting species of different lifespans. In: P.J. Fisheries and Aquatic Sciences 56, 97–104. Edwards, R.M. May and N.R. Webb (eds) Large-Scale Conservation and Exploitation 339

Ecology and Conservation Biology. Oxford: Blackwell McKinney, M.L. (1998) Is marine biodiversity at less risk? Scientific, pp. 199–227. Evidence and implications. Diversity and Distribu- Kokko, H., Lindström, J. and Ranta, E. (2001) Life histo- tions 4, 3–8. ries and sustainable harvesting. In: J.D. Reynolds, McKinney, M.L. (1999) High rates of extinction and G.M. Mace, K.H. Redford and J.G. Robinson (eds) Con- threat in poorly studied taxa. Conservation Biology 6, servation of Exploited Species. Cambridge: Cambridge 1273–81. University Press, pp. 301–22. Menge, B.A. (1995) Indirect effects in marine rocky Lande, R., Engen, S. and Saether, B.-E. (1995) Optimal intertidal interaction webs – patterns and importance. harvesting of fluctuating populations with a risk of Ecological Monographs 65, 21–74. extinction. American Naturalist 145, 728–45. Miller, R.R., Williams, J.D. and Williams, J.E. (1999) Lande, R., Engen, S. and Saether, B.-E. (2001) Sustain- Extinctions of North American fishes during the past able exploitation of fluctuating populations. In: J.D. decade. Fisheries 14, 22–38. Reynolds, G.M. Mace, K.H. Redford and J.G. Robinson Minkley, W.L. and Deacon, J.E. (eds) (1991) Battle (eds) Conservation of Exploited Species. Cambridge: Against Extinction: Native Fish Management in the Cambridge University Press, pp. 67–86. American West. Tucson, Ar: University of Arizona Lande, R., Saether, B.-E. and Engen, S. (1997) Threshold Press. harvesting for sustainability of fluctuating resources. Morris, A.V., Roberts, C.M. and Hawkins, J.P. (2000) The Ecology 78, 1341–50. threatened status of groupers (Epinephelinae). Biodi- Lau, P. and Parry-Jones R. (1999) The Hong Kong Trade in versity Conservation 9, 919–42. Live Reef Fish for Food. Hong Kong: TRAFFIC East Mosquera, I., Côté, I.M., Jennings, S. and Reynolds, J.D. Asia and World Wide Fund For Nature. (2000) Conservation benefits of marine reserves for Lewis, A.D., Adams, T.J.H. and Ledua, E. (1988) Fiji’s fish populations. Animal Conservation 3, 321–32. giant clam stocks – a review of their distribution, Moyle, P.B. and Leidy, R.A. (1992) Loss of biodiversity in abundance, exploitation and management. In: J.W. aquatic ecosystems: evidence from fish faunas. In: P.L. Copeland and J.S. Lucas (eds) Giant Clams in Asia and Fiedler and S.K. Jain (eds) Conservation Biology. New the Pacific. Australian Centre Monograph No. 9, York: Chapman & Hall, pp. 129–69. Canberra, Australia. pp. 66–72. Murawski, S.A., Brown, R., Lai, H.-L., Rago, P.J. and Liermann, M. and Hilborn, R. (1997) Depensation in fish Hendrickson, L. (2000) Large-scale closed areas as a stocks: a hierarchic Bayesian meta-analysis. Canadian fisheries-management tool in temperate marine sys- Journal of Fisheries and Aquatic Sciences 54, 1976– tems: the Georges Bank experience. Bulletin of Marine 84. Science 66, 775–98. Ludwig, D., Hilborn, R. and Walters, C. (1993) Uncertain- Musick, J.A. (1999a) Criteria to define extinction risk in ty, resource exploitation and conservation. Science marine fishes. Fisheries 24, 6–14. 260, 17–36. Musick, J.A. (1999b) Life in the slow lane: ecology and Mace, G.M. and Hudson, E.J. (1999) Attitudes towards conservation of long-lived marine animals. American sustainability and extinction. Conservation Biology Fisheries Society Symposium 23. Bethesda, Md: 13, 242–6. American Fisheries Society. Mackinson, S., Sumaila, U.R. and Pitcher, T.J. (1997) Nordeide, J.T. and Folstad, I. (2000) Is cod lekking or a Bioeconomics and catchability: fish and fishers’ promiscuous group spawner? Fish and Fisheries 1, behaviour during stock collapse. Fisheries Research 90–3. 31, 11–17. NRC (National Research Council) (1999) Sustaining Mangel, M. (2000) Irreducible uncertainties, sustainable Marine Fisheries. Washington, DC: National Acade- fisheries and marine reserves. Evolutionary Ecology my Press. Research 2, 547–57. NRC (National Research Council) (2000) Marine Pro- Matsuda, H., Takenaka, Y., Yahara, T. and Uozumi, Y. tected Areas: Tools for Sustaining Ocean Ecosystems. (1998) Extinction risk assessment of declining wild Washington, DC: National Academy Press. populations: the case of the southern bluefin tuna. Pace M.L., Cole J.J., Carpenter S.R. and Kitchell, J.F. Research in Population Ecology 40, 271–8. (1999) Trophic cascades revealed in diverse ecosys- McDowall, R.M. (1992) Particular problems for the con- tems. Trends in Ecology and Evolution 14, 483–8. servation of diadromous fishes. Aquatic Conservation Parent, S. and Schriml, L.M. (1995) A model for the Marine and Freshwater Ecosystems 2, 351–5. determination of fish species at risk based upon McDowall, R.M. (1996) Threatened fishes of the world: life-history traits and ecological data. Canadian Prototroctes oxyrhynchus Günther, 1870 (Prototroci- Journal of Fisheries and Aquatic Sciences 52, 1768– dae). Environmental Biology of Fishes 46, 60. 81. 340 Chapter 15

Pauly, D., Christensen, V., Dalsgaard, J., Froese, R. and exploitation. In: J.D. Reynolds, G.M. Mace, K.H. Red- Torres, F. Jr. (1998a) Fishing down marine food webs. ford and J.G. Robinson (eds) Conservation of Exploited Science 279, 860–3. Species. Cambridge: Cambridge University Press, pp. Pauly, D., Froese, R. and Christensen, V. (1998b) How per- 147–68. vasive is ‘fishing down marine food webs’?: response. Rijnsdorp, A.D., van Leeuwen, P.I., Daan, N. and Science 282, 1383. Heessen, J.L. (1996) Changes in abundance of demersal Petersen, C.W. and Levitan, D.R. (2001) The Allee effect: fish species in the North Sea between 1906–9 and a barrier to recovery by exploited species. In: J.D. 1990–5. ICES Journal of Marine Science 53, 1054–62. Reynolds, G.M. Mace, K.H. Redford and J.G. Robinson Roberts, C.M. (1997) Ecological advice for the global fish- (eds) Conservation of Exploited Species. Cambridge: eries crisis. Trends in Ecology and Evolution 12, 35–8. Cambridge University Press, pp. 281–300. Roberts, C.M. (2000) Why does fishery management so Pezzey, J.C.V., Roberts, C.M. and Urdal, B.T. (2000) A often fail? In: M. Huxham and D. Sumner (eds) Science simple bioeconomic model of a marine reserve. Eco- and Environmental Decision Making. London: logical Economics 33, 77–91. Prentice-Hall, pp. 170–92. Pinnegar, J.S.K., Polunin, N.V.C., Francour, P., Roberts, C.M. and Hawkins, J.P. (1999) Extinction risk in Badalamenti, F., Chemello, R., Harmelin-Vivien, M.L., the sea. Trends in Ecology and Evolution 14, 241–6. Hereu, B., Milazzo, M., Zabala, M., D’Anna, G. and Roberts, C.M. and Hawkins, J.P. (2000) Fully Protected Pipitone, C. (2000) Trophic cascades in benthic marine Marine Reserves: A Guide. Endangered Seas Cam- ecosystems: lessons for fisheries and protected- paign, WWF-US, Washington DC, and University of area management. Environmental Conservation 27, York, UK. 131pp. Available to download in English, 179–200. Spanish and French from: http://www.panda.org/ Pitcher, T.J. (1995) The impact of pelagic fish behaviour resources / publications / water / mpreserves / mar _ on fisheries. Scientia Marina 59, 295–306. dwnld.htm Polunin, N.V.C. and Roberts, C.M. (eds) (1996) Reef Roberts, C.M., Halpern, B., Palumbi, S.R. and Warner, Fisheries. Series in Fish Biology and Fisheries, No. 20. R.R. (2001). Designing networks of marine reserves: London: Chapman & Hall. why small, isolated protected areas are not enough. Pope, J.G., MacDonald, D.S., Daan, N., Reynolds, J.D. Conservation Biology in Practice. 2, 10–17. and Jennings, S. (2000) Gauging the vulnerability of Rodwell, L.D. and Roberts, C.M. (2000) Economic impli- non-target species to fishing. ICES Journal of Marine cations of fully-protected marine reserves for coral reef Science 57, 689–96. fisheries. In: H.S.J. Cesar (ed.) Collected Essays on the Powles, H., Bradford, M.J., Bradford, R.G., Doubleday, Economics of Coral Reefs. Sweden: CORDIO. W.G., Innes, S. and Levings, C.D. (2000) Assessing and Ruzzante, D.E., Wroblewski, J.S., Taggart, C.T., protecting endangered marine species. ICES Journal of Smedbol, R.K., Cook, D. and Goddard, S.V. (2000) Bay- Marine Science 57, 669–76. scale population structure in coastal Atlantic cod in Punt, A.E. (2000) Extinction of marine renewable Labrador and Newfoundland, Canada. Journal of Fish resources: a demographic analysis. Population Ecology Biology 56, 431–47. 42, 19–27. Sadovy, Y. (2001) The threat of fishing to highly fecund Reaser, J.K., Pomerance, R. and Thomas, P.O. (2000) fishes. Journal of Fish Biology 59 (Supplement A), Coral bleaching and global climate change: scientific 90–108. findings and policy recommendations. Conservation Sadovy, Y. and Eklund, A.-M. (1999) Synopsis of bio- Biology 14, 1500–11. logical data on the Nassau grouper, Epinephelus stria- Reynolds, J.D. and Jennings, S. (2000) The role of tus (Bloch, 1792) and the Jewfish, E. itajara animal behaviour in marine conservation. In: (Lichtenstein, 1822). U.S. Dep. Commer., NOAA L.M. Gosling and W.J. Sutherland (eds) Behaviour Technical Report NMFS 146, and FAO Fisheries Syn- and Conservation. Cambridge: Cambridge University opsis 157, 65pp. Press, pp. 138–257. Sadovy, Y. and Leung, C.-W. (2001) The case of the Reynolds, J.D. and Mace, G.M. (1999) Risk assessments disappearing croaker, the Chinese bahaba, Bahaba of threatened species. Trends in Ecology and Evolution taipingensis. Porcupine 24, 14–15. 14, 215–17. Scheffer, M., Carpenter, S., Foley, J.A. and Walker, B. Reynolds, J.D, Mace, G.M., Redford, K.H. and Robinson, (2001) Catastrophic shifts in ecosystems. Nature 413, J.G. (eds) (2001a) Conservation of Exploited Species. 591–6. Cambridge: Cambridge University Press. Smith, D.C., Fenton, G.E., Robertson, S.G. and Short, Reynolds, J.D., Jennings, S. and Dulvy, N.K. (2001b) Life S.A. (1995) Age determination and growth of orange histories of fishes and population responses to roughy (Hoplostethus atlanticus) – a comparison of Conservation and Exploitation 341

annulus counts with radiometric ageing. Canadian Sea between 1930 and the present day. ICES Journal of Journal of Fisheries and Aquatic Sciences 52, 391– Marine Science 55, 392–402. 401. Watling, L. and Norse, E.A. (1998) Disturbance of the Smith, F.D.M., May, R.M., Pellew, R., Johnston, T.H. and seabed by mobile fishing gear: a comparison to forest Walter, K.R. (1993) How much do we know about the clearing. Conservation Biology 12, 1180–97. current extinction rate? Trends in Ecology and Evolu- Watts, J. (2001) Loadsa tunny £570/kg price record. tion 8, 375–8. Guardian, London (6 January 2001) p. 16. Smith-Vaniz, W.F., Colette, B.B. and Luckhurst, B.E. Welcomme, R.L. (1988) International Introduction of In- (1999) Fishes of Bermuda. American Society of Ichthy- land Aquatic Species. FAO Fisheries Technical Paper, ology and Herpotology, Special Publication No. 4. No. 294. Rome: FAO. Lawrence, Kansas: Allen Press. Wilkinson, C.R. (2000) The Status of Coral Reefs of the Spalding, M.D. (1998) ‘Patterns of biodiversity in coral World: 2000. Australian Institute of Marine Sciences, reefs and mangrove forests: global and local scales’. Global Coral Reef Monitoring Network, Townsville, PhD thesis, University of Cambridge, UK. Queensland. Stevens, J.D., Bonfil, R., Dulvy, N.K. and Walker, P. (2000) Wing, S.W. (2001) The sustainability of resources used by The effects of fishing on sharks, rays and chimaeras native Americans on four Caribbean Islands. Interna- (chondrichthyans), and the implications for marine tional Journal of Osteoarchaeology 11, 112–26. ecosystems. ICES Journal of Marine Science 57, Witte, F., Goldschmidt, T., Goudswaard, P.C., Ligtvoet, 476–94. W., van Oijen, M.J.P. and Wanink, J.H. (1992) Species Stobutzki, I., Miller, M. and Brewer, D. (2001) Sustain- extinction and concomitant ecological changes in ability of fishery bycatch: a process for assessing highly Lake Victoria. Netherlands Journal of Zoology 42, diverse and numerous bycatch. Environmental Con- 214–32. servation 28, 167–81. Wolff, W.J. (2000) Causes of extirpations in the Wadden Strong, D.R. (1992) Are trophic cascades all wet – differ- Sea, an estuarine area in the Netherlands. Conserva- entiation and donor-control in speciose ecosystems. tion Biology 14, 876–85. Ecology 73, 747–54. Yoklavich, M.M. (1998) Marine harvest refugia for Wabnitz, C. and Polunin, N.V.C. (2001) The use of marine west coast rockfish: a workshop. NOAA Technical reserves to protect fish stocks. Fisheries Society of the Memorandum NMFS-SWFSC-255, Silver Springs, British Isles Position Paper No. 1. Maryland. Walker, P.A. and Hislop, J.R.G. (1998) Sensitive skates or Zann, L. (1992) The State of the Marine Environment in resilient rays? Spatial and temporal shifts in ray species Fiji. Townsville: National Environmental Manage- composition in the central and north-western North ment Project. 16 Ecosystem Effects of Fishing

M.J. KAISER AND S. JENNINGS

16.1 INTRODUCTION species may be predators, grazers, prey, scavengers and competitors. Some species may form habitat Humans are extremely effective predators. As a structures (Lenihan 1999) or maintain habitat result we have seen at various times throughout patchiness through their feeding activities (Van history the sequential overexploitation of many Blaricom 1982). The consequences of harvesting a target species (FAO 1994). Traditional fisheries particular species will depend on its role and domi- management focused on these single species with nance within an ecosystem and also the com- a recent move towards multispecies approaches. plexity of the species interactions in that system It has proved difficult enough to manage stocks of (Pauly and Christensen, Chapter 10, this volume). single species successfully, and so it is not difficult Hence intensive harvesting of one species might to understand why relatively little effort has been relieve predation on its prey species while harvest- devoted to trying to predict ecosystem responses ing prey species might increase predation pressure to target-species removal. Nevertheless, the wider on alternative prey types. When they exist, such ecosystem effects of fishing activities are becom- ecosystem responses to single-species removal ing more apparent, which emphasizes our need to are known as ‘trophic cascades’ (Pace et al. 1999). understand the consequences of exploiting marine Later in this chapter we give some examples where resources (Hall 1998; Jennings and Kaiser 1998; trophic cascades have resulted from fishing activi- Auster and Langton 1999). Although we reviewed ties. However, a recent study by Yodzis (1998) this topic in detail several years ago (Jennings and demonstrated that even when dealing with a well- Kaiser 1998), there have been new and exciting de- quantified ecosytem such as the Benguela system, velopments in the interim period, some of which the predicted outcome of culling large top pre- have confirmed our previously unsubstantiated dators such as seals can be highly variable (see theories. Section 16.5). While intensive fishing has led to well- documented population reductions of many target species (e.g. Davis et al. 1996; Myers et al. 1996), it 16.2 EFFECTS OF has also caused changes in the species composi- HARVESTING tion structure within fish communities (e.g. TARGET SPECIES Greenstreet and Hall 1996). Whereas in the past fish assemblages had a broad range of size classes, A target species is just one component of an exploited assemblages tend to be dominated by a ecosystem; hence it may perform a variety of much higher proportion of small-sized individuals functions as it interacts with other species. Target (Kirkwood et al. 1994; Pauly et al. 1998). Jennings Ecosystem Effects of Fishing 343 et al. (1998) used a comparative approach based on (Carangidae) declined in abundance rapidly as fish- phylogenetic comparisons to examine the differ- ing pressure increased and recovered slowly when ential effects of fishing on individual species that fishing ceased. During periods when the commu- have contrasting life histories. They found that nity was most intensively exploited, they formed a those fishes that decreased in abundance relative smaller proportion of the total abundance of the to their nearest phylogenetic relative matured community and the community was dominated later and at a greater size, grew more slowly to- by smaller species with faster life histories (see wards a greater maximum size and had lower also Polunin, Chapter 14, this volume). rates of potential population increase (Table 16.1). Thus the response of fish populations to ex- For example, common skate (Raja batis) have ploitation follows a consistent pattern, with the the steepest slope for the decline in catch rates progressive removal of the largest body-sized fauna through time, whereas the smaller starry rays from the system leading to dominance by smaller- (Raja radiata) actually show an increase in popula- bodied individuals. Consequently, the sustained tion numbers. While both halibut (Hippoglossus harvesting of fish populations has led to a gradual hippoglossus) and long rough dabs Hippoglos- decline in their body-size spectrum (Rice and soides platessoides have decreased with time, Gislason 1996). This has important considerations the rate of decrease is three times greater for the for predator–prey interactions. For example, juve- halibut that attains a much greater maximum nile cod, Gadus morhua, are preyed upon by adult size. whiting, Merlangius merlangus, while the latter Patterns observed in these North Sea studies are consumed by adult cod. A decrease in the body- are also typical of those in exploited reef fish size spectrum of cod populations could mean that communities. Russ and Alcala (1998a,b) looked at they are exposed to proportionately greater levels changes in a Philippine reef fish community dur- of predation. Conservation measures designed ing periods of exploitation and protection from to reduce bycatches of whiting in the cod fishery fishing. Large predatory species with slow life could further exacerbate this problem by further histories, such as snappers (Lutjanidae), emperors reducing the body-size spectrum of cod through (Lethrinidae), sea basses (Serranidae) and jacks harvesting while conserving the population struc-

Table 16.1 Abundance trends and life-history parameters for selected species in the North Sea fish community.

-1 Species Common name Trend L• (cm) K ( y ) Tm (y) Lm (cm)

Raja batis Common skate -0.030 254 0.06 11.0 130 Raja naevus Cuckoo ray -0.027 92 0.11 9.0 59 Raja radiata Starry ray +0.013 66 0.23 4.0 46 Squalus acanthias Spurdog -0.051 90 0.15 6.5 67 Scyliorhinus canicula Lesser-spotted dogfish +0.002 88 0.20 5.0 58 Trisopterus minutus Poor cod -0.007 20 0.51 2.0 15 Trisopterus esmarkii Norway pout +0.012 23 0.52 2.3 19 Hippoglossus hippoglossus Halibut -0.015 204 0.10 5.8 83 Hippoglossoides platessoides Long rough dab -0.005 25 0.34 2.6 15

Notes: Trend: slope of linear relationship between standardized catch rate (numbers h-1) in standard fisheries sur- veys and time (years); L•: asymptotic (maximum) length; K: growth rate; Tm: age at maturity; Lm: length at maturity. The sequence and grouping of species reflects their phylogenetic relationship. The species are shown in phylogeneti- cally linked couplets or triplets and the first species is the one for which catch rate has declined the most. (Source: adapted from Jennings et al. 1999.) 344 Chapter 16 ture of whiting through technical conservation Urchins erode the reef matrix as they graze and measures. this prevents the settlement and growth of coral re- In the next few sections we give some examples cruits. Occasionally recruitment failure or urchin of where cascade effects have and have not been disease may lead to a collapse of urchin popula- observed in marine systems in response to the tions. In the absence of such occurrences, inter- selective removal of target species. vention may be required to promote recovery of the reef ecosystem by the deliberate removal of 16.2.1 Exploitation of target urchins. McClanahan et al. (1996) found that when species on tropical reefs they removed urchins from unfished experimental plots on Kenyan reefs, there were significant in- Most tropical reefs are fished intensively since creases in algal cover and fish abundance within they provide the main sources of protein and in- one year. However, on fished reefs, herbivorous come for fishers who have few other means of gen- fishes were less abundant, and the algae rapidly erating a living. Many trophic groups are targeted, overgrew corals as they proliferated in the absence and the abundance of herbivorous, piscivorous or of herbivores. invertebrate feeding fishes is often reduced by an While many studies have shown that the abun- order of magnitude or more on fished reefs (Russ dance of piscivorous reef fishes is greatly reduced 1991). by fishing, there is little evidence for a correspond- Fishes and sea urchins are the most abundant ing increase in the abundance of the fish species herbivores on most tropical reefs. Sea urchin abun- that are their prey (Bohnsack 1982; Russ 1985; dance is regulated by recruitment success, food Jennings and Polunin 1997). Why is the response supply and natural mortality due to predation by of prey fish communities so weak? The reasons for fish and disease. The main predators of sea urchins this are probably linked to reef fish community are invertebrate feeding fishes such as emperors structure, in which phylogenetic groupings of fish (Lethrinidae) and triggerfishes (Balistidae) and, contain many species, with a wide range of life on reefs in the Caribbean and East Africa, fish history traits, behavioural differences and feeding predation strongly regulates urchin populations strategies (Hiatt and Strasburg 1960; Parrish et al. (McClanahan 1995a). As a result, on reefs where 1985, 1986; Hixon 1991). Moreover, most fish populations of emperors and triggerfishes have species alter their feeding behaviour and hence been reduced by fishing, there is good evidence diet as they grow and can act as both prey and that urchins have proliferated (McClanahan and predators of other species. Hence, while the collec- Muthiga 1988; McClanahan 1992, 1995b). tive impacts of predatory fish are large, the impacts When urchins dominate herbivore biomass, of individual predator species on the dynamics of they will graze the majority of algal production on their prey are minor (Hixon 1991). a reef. Because urchins have low consumption and Notably, at smaller scales (m2 compared with respiration rates they can survive and reproduce km2) there is some evidence for the role of pre- when their food supply is greatly reduced by their dation as a structuring force, particularly when own feeding activities. In contrast, herbivorous habitat or refuge space is directly limited. Thus fishes have higher consumption and respiration experimental reductions in piscivore abundance rates and are unable to tolerate low levels of food led to detectable decreases in the abundance and availability. As a result, urchins outcompete her- diversity of their prey (Caley 1993; Hixon and bivorous fishes and reach maximum biomass Beets 1993; Carr and Hixon 1995). levels an order of magnitude higher (McClanahan 1992). Since the herbivorous fishes are poor com- 16.2.2 Removal of predatory fish in petitors, they may be unable to achieve their temperate marine fisheries former abundance when fishing is stopped (McClanahan 1995a). Even in the most intensively fished marine ecosys- Ecosystem Effects of Fishing 345 tems such as the Benguela upwelling, Georges cascading trophic effects is much more prevalent Bank, Bering Sea and North Sea, piscivorous fish following the exploitation of target fishes in fresh- are more important consumers of fish biomass water lakes and streams (Marten 1979; Carpenter than the combined removal by humans, marine et al. 1987; Persson, Chapter 15, Volume 1). For ex- mammals and birds (e.g. Bax 1991; Yodzis 1998). ample, Flecker and Townsend (1994) and McIntosh Long-term studies have demonstrated that the and Townsend (1996) demonstrated that the pres- intensive removal of predators has allowed other ence of a brown trout (Salmo trutta) predator in a species to proliferate. One such example showed New Zealand stream led to increased predation of that population explosions of sandeels (Am- herbivorous mayfly larvae. The mayfly larvae re- modytes spp.) in the northern Atlantic coincided sponded by spending more time sheltering beneath with the depletion of predators such as the herring rocks and less time feeding on their upper surfaces. (Clupea harengus) and mackerel (Scomber scom- This change in mayfly larval behaviour led to a brus) (Sherman et al. 1981). Although coincident proliferation of epilithic algae. Conservation or changes in predator and prey populations might fisheries management action that affects the trout infer causality, it is just as likely that such changes population would therefore have knock-on effects could be driven by a range of environmental on other aspects of ecosystem function. factors. Carpenter et al. (1987) examined the possibil- Elsewhere, we have concluded that there are ity that species at higher trophic levels could regu- relatively few circumstances in which changes in late the abundance, productivity and community the abundance of piscivorous fishes in marine structure of phytoplankton and zooplankton. In ecosystems has had cascading impacts on other order to do this, they manipulated the fish popula- parts of the system (Jennings and Kaiser 1998). tions of three entire lakes. Productivity varied Most fish species can act as both predators and prey annually in the control lake, predominantly in in the course of their life history and adult pre- response to climatic factors. The addition of dators are capable of switching diet and their piscivorous largemouth bass (Micropterus feeding behaviour in response to prey availability salmoides) and hence potentially the removal of (Mittlebach, Chapter 11, Volume 1). Hence high planktivorous fish from one of the lakes, led to an fish consumption rates do not necessarily imply increase in zooplankton biomass and a shift from a that predation is a structuring force of fish popula- copepod and rotifer zooplankton assemblage to tions within a particular system. Indeed, in tem- one dominated by cladocerans. As a result, algal perate systems, the strongest evidence for the biomass decreased and hence primary productiv- predator-based control of prey species comes from ity also fell. In the lake where most largemouth the impact of humans on their target species. The bass were removed, thereby relieving the preda- strength of this relationship is likely to result from tion pressure on the planktivores, zooplankton the conservative fishing strategies employed by biomass and the dominance of large cladocerans humans. In the majority of commercial fisheries, increased. At first this seems surprising, but it fishers are unwilling to be flexible in their aims and appeared that the planktivorous fish remained target a relatively small proportion of the total fish within refuges despite lower predator numbers, fauna. Most predatory fish, conversely, are very gen- and thus allowed the larger zooplankton to in- eralist feeders, often switching to invertebrate prey crease. Carpenter et al. (1987) had thus provided or cannibalism and eating many species of fishes at convincing evidence that fish populations could be different stages in their life history (Fig. 16.1). important regulators of primary productivity and that predator avoidance behaviour also has impor- 16.2.3 A comparison with tant consequences for other trophic levels in a freshwater systems system. Unlike in many marine systems, the occurrence of 346 Chapter 16

Birds

South African fur seals

Sharks Cetaceans

Other pelgics Tunas

Hakes Snoek

Horse mackerel Chub mackerel

Kob

Round herring Anchovy Pilchard

Lightfish Goby

Other groundfish Lanternfish Geelbeck

Squid

Macrozooplankton Benthic Benthic filterfeeders carnivores Mesozooplankton Gelatinous zooplankton Bacteria Phytoplankton Microzooplankton Detritus

Fig. 16.1 Schematic diagram of the Benguela food web adapted from Yodzis (1998). Many of the linkages and some of the species have been removed for clarity. What is clear is that there are many linkages between the different trophic levels in the system and cannibalism (indicated by the circular arrows) occurs often.

16.2.4 Why systems respond and Kaiser 1998). Within lake ecosystems most differently to predator removal biomass is aggregated into a few size classes. These loosely correspond to phylogenetic groupings such The removal of fish predators in marine ecosys- as algae, crustaceans or fish with a limited range tems often has limited consequences for prey in their biomass spectrum (Sprules et al. 1983; species, whereas in many freshwater ecosystems Sprules and Goyke 1994; Sprules and Stockwell there are much closer links between predator and 1995). Conversely, in the marine environment the prey dynamics (Persson, Chapter 15, Volume 1). biomass spectrum is often extended and there is We attribute this to general differences in the more variance in size within the main phylogene- structure of many marine and freshwater ecosys- tic groupings. Moreover, phylogenetic groupings tems and suggest that the indirect effects of fish tend to contain more species, with a wider range of predator removal in marine systems are smaller life history traits, behaviours and feeding strate- and less predictable than in freshwaters (Jennings gies (Jennings and Kaiser 1998). Ecosystem Effects of Fishing 347

16.2.5 The collective effects of fishing larval retention of bluehead wrasse (Thalassoma bifasciatum) at St Croix (US Virgin Islands) in the Designing experiments to study fishing effects Caribbean. They used isotopic signatures in the has always been difficult because it is never clear otoliths to identify the source of larvae that were whether areas subject to different fishing pressure recruiting to the reefs, and showed that many had can be treated as replicates that respond more or been retained close to the area where they were less independently to fishing. Spatial designs have spawned. Jones et al. (1999) marked the otoliths often been used to investigate the effects of fishing of approximately 10 million developing Ambon in the tropics, with reefs or islands treated as repli- damselfish (Pomacentrus amboinensis) eggs with cates. This approach was adopted by Samoilys tetracycline, a marker that fluoresces when the (1988) and Jennings and Polunin (1996) to examine otolith is examined under ultraviolet light. All the effects of fishing on the structure of fish com- the eggs were marked on a reef at Lizard Island on munities, and by McClanahan to look at the fish, the Great Barrier Reef, and they comprised 0.5– algal and urchin interactions described previously. 2.5% of total Ambon damselfish egg production The assumption that different reefs could be on this reef. Later in the year, juveniles recruiting treated as semi-independent replicates was to the reef were caught, and their otoliths were thought to be reasonable because small (km2) areas removed and viewed under ultraviolet light. The protected from fishing, such as marine reserves, frequency of otoliths with marks suggested that would harbour a diverse and abundant fish com- 15–60% of juveniles returned to the natal reef and munity when adjacent reefs were heavily fished. In that long-distance dispersal was not usual. addition, fish communities on reefs that were sep- Thus there is some evidence to suggest that arated by a few km would respond differently to fish populations are largely self-sustaining on reefs local fishing pressure. It was also assumed that the that are separated by >10km. This suggests that migrations or movements of adult reef-associated spatial designs can be useful for looking at long- fishes between replicate areas were limited, and term fishing effects and that even small and iso- this was later confirmed with tagging studies (e.g. lated marine reserves may be useful conservation Holland et al. 1993; Zeller and Russ 1998). How- tools for some reef fish species. (See also Polunin, ever, the major weakness in understanding the Chapter 14, this volume, for further commentary level of connectivity between reefs was associated on the degree to which fish move away from small with the planktonic egg and larval stages (Jones et areas.) al., Chapter 16, Volume 1; Polunin, Chapter 14, this volume). These could return to their natal reef, making the effects of fishing localized, or be 16.3 EFFECTS OF PREY transported to reefs that were many tens to hun- REMOVAL ON SEA BIRDS dreds of km away, making the effects of fishing col- lective. Clearly, if the majority of larvae recuited to Piscivorous sea birds feed primarily on small sites away from the natal reef, then spatial studies shoaling fish. In the North Atlantic, these prey of fishing effects gave little insight into the longer- species include herring (Clupea harengus), sprat term impacts of fishing. Moreover, if we do not (Sprattus sprattus) and sandeels (Ammodytes know the scale on which we should test for fishing spp.). Sandeels are particularly important prey dur- effects, it is equally difficult to know the scale on ing the breeding and chick-rearing season, whereas which to impose management strategies. clupeoid and gadoid fish become more important Two recent studies made direct assessments of dietary components during the winter period. larval dispersal rates. Both concluded that rates The International Council for the Exploration of of dispersal between reefs were low and that the Sea (ICES) working group on sea bird ecology most larvae were retained close to the natal reef. (Anon. 1994) estimated that the annual consump- Swearer et al. (1999) looked at the evidence for tion of fish by 18 sea bird species, based on sea bird 348 Chapter 16

250000

200000

150000

100000 Tonnes per year

50000

0 Fig. 16.2 Estimated total annual consumption of fish and discarded Offal Sandeels Gadoids Discards materials by 18 sea bird species in Mackerel Other prey Large herring the North Sea. (Source: data from Sprat and Herring Anon. 1994.) abundance estimates and their energetic require- 16.3.1 Effects of industrial fisheries -1 ments was 600000 tonnes y (Fig. 16.2). Of this on sea bird populations figure the total consumption of small fish, con- sisting of sandeel, sprat, young herring and small Industrial fisheries in the North Sea have in- gadoids, by sea birds is approximately 250000 creased substantially since the early 1960s. These tonnes y-1, which is about one-tenth of the small fisheries target species such as sandeels, which fish consumed by larger predatory fish (Anon. are then converted into animal feed, fertilizer or 1994). are even used as fuel. This fishery has caused It is thought that the abundance of immature concern among conservationists because many fish in the North Sea has increased due to the over- top predators such as seals, cetaceans and sea birds, fishing of larger size classes of piscivorous fish (e.g. rely on sandeels as the main part of their diet Furness and Monaghan 1987; Daan et al. 1990). (Furness 1990). The direct impacts of fisheries im- Similarly, congruent shifts in sandeel abundance pacts on prey availability can easily be overlooked in the western and eastern North Atlantic ecosys- as sea birds are able to compensate for food short- tems were also explained by predator removal by ages by working harder, allocating more time to fisheries (Sherman et al. 1981). Although the in- foraging, switching to alternative prey types, or crease in the amount of small fish intuitively must abandoning nesting attempts. Ultimately the im- have resulted in a larger potential resource for pis- pacts of significant reductions in prey popula- civorous sea birds, the availability of prey may not tions become apparent as sea birds starve to death. have changed quite that dramatically. Shifts in the During the 1980s, the breeding success of sev- localized availability of small fish may negatively eral sea bird species in the Shetland Islands de- affect sea birds even when the wider fish stock is clined markedly, coincident with a marked decline relatively unaffected. For example, the reproduc- in landings of sandeels from an industrial fishery tive output of puffins (Fratercula arctica) that nest that operated close to the Shetland coast (Mon- on the Lofoten islands off Norway was reduced aghan et al. 1992). The sea bird species affected over a 20-year period following the collapse of local most severely were surface feeders such as Arctic herring stocks, even though, elsewhere, herring terns (Sterna paradisaea), black-legged kittiwakes stocks remained at relatively high levels (Anker- (Rissa tridactyla) and great skuas (Catharacta Nilssen 1992). skua) that fed predominantly on O-group sandeels. Ecosystem Effects of Fishing 349

The kleptoparasitic Arctic skuas (Stercorarius gregations of salmon are important sources of food parasiticus) were also affected, as was the puffin. for many terrestrial animals and bird species that The diet of gannets changed from predominantly inhabit the relevant watersheds. Studies that have sandeels to clupeoid and gadoid fish. After a few examined 15N/14N ratios in food webs of systems years of poor breeding success, significant declines with anadromous salmon and those without indi- occurred in breeding populations of Arctic terns, cate that the decomposing carcasses of salmon black-legged kittiwakes and common guillemots. are remineralized by primary producers that are The numbers of Arctic terns increased dramatical- consumed by fish predators (Kline et al. 1993; ly again in 1991, coincident with the appearance of Cederholm et al. 1999). a large year class in the sandeel population. Salmon carcasses can supply a critical source These population shifts led to suggestions that of energy for some terrestrial vertebrates and pro- the industrial fishery competed for the same re- vide nutrients for riparian vegetation along some source as the sea birds and was responsible for the spawning streams. They are now viewed by some decline in their food supply. However, other stud- as keystone species in certain terrestrial vertebrate ies indicated that the decline in sandeel abundance communities (Willson and Halupka 1995). The was the result of poor recruitment to the Shetland reproduction cycle and seasonal distribution of stock. Hence, natural fluctuations in sandeel sur- species such as bald eagles (Haliaeetus leuco- vival and recruitment may have been the main cephalus) is tied directly to a spawning run of cause of the decline in prey availability for sea salmon (McClelland et al. 1982). Salmon stocks birds breeding on the Shetland Islands (Anon. are, in general, overexploited (Nehlsen et al. 1991). 1994). Hence the present-day inputs of marine nutrients from salmon carcasses are probably much lower than they were prior to overexploitation. While it 16.4 HOW FISHERIES is not possible to attribute differences in past and AFFECT ENERGY SUBSIDIES present-day community structure to decreases in 16.4.1 Terrestrial systems the amount of energy and nutrient subsidies pass- ing into watershed systems, the importance of the At first glance, the ecosystem consequences of role of salmon carcasses has prompted manage- fishing in the marine environment might seem to ment action to ensure their continued presence. have little relevance to ecosystem processes in ter- For example, timber debris on which salmon car- restrial ecosystems. However, there is a substan- casses become snagged is introduced into river tial literature that has examined the importance of systems to avoid loss of the carcasses through the contribution of nutrients and energy to aquatic wash-out, and salmon carcasses from fish farms and terrestrial ecosystems by salmon carcasses are used to supplement nutrient inputs in some (See also Persson, Chapter 15, Volume 1, for a dis- systems (Cederholm et al. 1999). cussion of the role of fishes in nutrient cycling within freshwater systems.). Pacific salmon (On- corhynchus spp.) migrate to sea where they grow 16.4.2 Energy subsidies from to maturity before migrating back to their native discarding practices rivers to spawn and in many cases die (Cederholm et al. 1999). Thus Pacific salmon are important While some fishing techniques are highly specific, vectors of marine-derived nutrients that are such as pelagic fisheries and single hook and moved upstream. Energy subsidies, such as those line fisheries, others yield bycatches or incidental provided by salmon carcasses, are only likely to catches of species that are not utilized by the have a significant ecological contribution if they fishers. These bycatches are usually discarded and occur frequently enough and in large concentra- are composed of animals that are dead, dying or in- tions (Polis and Strong 1996). Mass spawning ag- jured (Kaiser and Spencer 1995). Bycatches may be 350 Chapter 16 discarded for legal reasons, in the case of commer- consumed by sea birds each year and this partly ex- cially valuable species that are undersized. Alter- plains the increases in the populations and distrib- natively, bycatches are discarded because there is ution of some species (Furness 1996; Camphuysen insufficient money to be gained by sorting or land- and Garthe 2000). Sea birds are very choosy, prefer- ing them. There are even instances when fishers ring to eat offal that gives the greatest energetic will discard commercial species that are greater reward in preference to flatfishes or invertebrates than the minimum landing size so as to enable that yield lower energy returns or that are more them to fill their quota with a larger average size of difficult to swallow. Relatively high consumption fish. This phenomenon is known as high-grading rates for these less-preferred discard types occurs (Pascoe 2000). Global energy subsidies in the form in areas of the sea where the resource needs of the of bycatch material dumped back into the ocean sea birds at sea exceeds the amount of material dis- totals approximately 27 million tonnes per annum carded (Table 16.2). This suggests that competition which is almost equivalent to a quarter of the for fisheries discards is intense in some areas of the global annual landings of fish (Alverson et al. sea, but not in others. 1994). However, this figure does not account for The northern fulmar (Fulmarus glacialis) is one those animals that are never landed on board ves- of the most common scavenging sea birds that sels but remain dead on the seabed. Thus, fishing are observed following fishing vessels in the North activities redirect energy in marine ecosystems Sea, and has shown the most spectacular increase by the production of carrion that then becomes in numbers (Fig. 16.3) and breeding range of any sea available to scavengers on the seabed and those bird found in the North Atlantic (Lloyd et al. 1991). that feed at the surface of the sea. The expansion of the fulmar population has been attributed mainly to its success as an offal scav- 16.4.3 Changes in sea enger, firstly associated with whalers in northern bird populations polar regions, and later in association with com- mercial fishing fleets. Field studies have ranked Anyone who has witnessed a fishing boat return- fulmars at the apex of the interspecific dominance ing to port will have observed the scrum of sea hierarchy, hence they are able to obtain the best birds that contest for the offal and discarded fish as quality food: offal or fish livers (Hudson and they are thrown overboard. In the North Sea alone, Furness 1988). While Tasker et al. (1987) conclud- approximately 235000 tonnes of discarded fish are ed that fishing activities were an important deter-

Table 16.2 The energetic requirements of sea birds at sea and the energetic content of available discards. (Source: from Camphuysen and Garthe 2000.)

Subregion of the Energetic Energetic equivalents Index of Potential shortage North Sea requirements of of discards and offal required/available or surplus if scavengers (¥106 kJ) discards discards were only (¥106 kJ) prey

Northwestern 899000 463000 1.9 shortage Northeastern 312000 456000 0.7 surplus Central-western 364000 269000 1.4 shortage Central 322000 659000 0.5 surplus Central-eastern 120000 698000 0.2 surplus Southern 175000 871000 0.2 surplus Total 2192000 3216000 0.6 surplus Ecosystem Effects of Fishing 351

800000

600000 All savenging sea birds

400000

Numbers Fulmar

200000 Fig. 16.3 Increases in the num- Kittiwake bers of breeding pairs of a selection of scavenging sea birds breeding on Gannet North Sea coasts. (Source: adapted 0 from Camphuysen and Garthe 1900 1920 1940 1960 1980 2000 2000.) Year

minant of fulmar distribution at sea, more recent Further investigations indicated that fulmars had studies, conducted at different scales, were less a northerly, offshore distribution, over waters with conclusive and suggested that, although fulmars a rather strong Atlantic influence and high con- were evidently attracted by fishing vessels, the centrations of zooplankton. This observation sug- spatial distribution of the main fisheries and gested that the distribution of zooplankton, their fulmars did not coincide as well as was expected natural prey, was a much more important parame- (Camphuysen et al. 1995; Stone et al. 1995). Also, ter that influences their distribution at sea. recent investigations have shown that the position While offal and discards would appear to pre- in the dominance hierarchy at fishing vessels was sent sea birds with high-energy food that is easily not so high in other parts of the North Sea, as com- obtained, this does not always equate to higher re- pared with studies that had concentrated on the productive output as might be expected. Hamer et Shetland Islands (Camphuysen et al. 1993). al. (1991) demonstrated that the chick growth In light of these findings, Camphuysen and index of great skuas on Foula (Shetland Islands) de- Garthe (1997) re-examined the distribution and clined considerably when more than half of the scavenging habits of northern fulmars in the prey delivered by the parents was comprised of North Sea using a large set of recently gathered discards and offal. When smaller proportions data. They calculated the availability of discards (20–30%) of offal and discards were included in and offal per km2 for each subregion in the North chick diets, growth rates were relatively high. Sea (Table 16.2). They found negative correla- The insights into the dominance hierarchies tions for northern fulmar densities at sea versus that exist between different scavenging sea birds the availability of offal and all types of fisheries have important implications for fisheries man- wastes. There was no correlation between num- agers as they suggest that measures to reduce the bers of fishing vessels at sea and densities of amount of discards would affect the lower ranked northern fulmars at this scale, but the numbers sea bird species first. Increases in trawl-net mesh of fulmars observed at the stern of fishing vessels size would increase the mean length of discarded were highly correlated with the overall densities of fish and this in turn would lower the handling northern fulmars at sea. These results suggested efficiency of smaller scavenging sea birds that are strongly that commercial fisheries were not the less well adapted to consuming large fish (Furness, prime determinant of fulmar distribution at sea. Ensor and Hudson 1992). A total ban of discards 352 Chapter 16 and offal would lead to major changes in current sea bird diets. An overnight introduction of such a (a) ban would certainly increase the predation risk for smaller birds from larger predators such as great skuas (Camphuysen and Garthe 2000).

16.4.4 Energy subsidies for marine benthic communities Food-falls of carrion can have a profound effect on local diversity and production processes of deep- sea benthic communities (Dayton and Hessler 1972). Although fishing activities occur in oceanic waters, the majority of fishing activities are con- (b) fined to more shallow coastal waters that are less than 100m deep. Hence the energy subsidies to deep-sea communities from current levels of fish- ing activities are likely to be less significant than for shallow waters. The greatest energy subsidies to benthic com- munities are generated by towed bottom fishing gears. These trawls and dredges are specifically de- signed to remain in close contact with the seabed and thereby maximize catches of bottom-dwelling species such as flatfishes and scallops (Misund et al., Chapter 2, this volume). Not surprisingly, these gears are particularly heavy and robust; a typ- Fig. 16.4 Towed bottom fishing gears are designed to ical beam trawl fished in the North Sea will weigh remain in close contact with the seabed: (a) 12-m wide commercial Dutch beam trawls can be fitted with up to up to 10 tonnes in air. Most bottom fishing gears 27 tickler chains that are attached between the two are fitted with chains or toothed bars across the beam shoes, (b) scallop dredges are fitted with toothed mouth of the net or dredge (Fig. 16.4; see also bars that dig scallops out of the seabed. Misund et al. Chapter 2, this volume). These modi- fications are designed to dig out or disturb the tar- get species from the substratum whereupon they areas of trawl disturbance often exceeds their are caught in the following net. Although these initial density before trawling started by between fishing gears are very effective at catching the tar- 4 and 10 times (Ramsay et al. 1996; Fonds and get species (Cruetzberg et al. 1987), they also catch, Groenwold 2000). Examination of the gut contents damage or kill in situ a large number of non-target of scavenging fish reveals that they eat a wider fish and invertebrates (Bergman and Van Santbrink variety and larger rations of prey in areas of trawl 2000). The odours emitted from this carrion rap- disturbance (Kaiser and Spencer 1994; Kaiser and idly attract mobile scavenging species into areas Ramsay 1997). Furthermore, they are able to con- of the seabed that have been trawled (Kaiser and sume prey such as burrowing sea urchins and large Spencer 1994; Ramsay et al. 1996, 1997). Fishers bivalve molluscs that normally would be unavail- have known of this phenomenon for many years able to them (Kaiser & Spencer 1994). In addition and it is not uncommon to see trawlers fishing one to scavenging fishes, a wide variety of invertebrate after the other along the same navigational plot. scavengers are attracted into areas of trawl distur- The density of scavenging fishes attracted into the bance. The most commonly observed taxa that Ecosystem Effects of Fishing 353 feed upon fisheries carrion are starfishes (Aster- reliable and constant source of high-quality food oidea), hermit crabs (Paguridae), crabs (Decapoda) (Camphuysen et al. 1993). Furthermore, we now and gastropods (Gastropoda) (Hill and Wassenberg know that sea birds will often choose not to feed on 1990; Kaiser and Spencer 1996a; Ramsay et al. fisheries discards especially when rearing chicks 1997). When hermit crabs are found in high back- (Camphuysen and Garthe 2000). In contrast, ben- ground densities, they can locate carrion within thic invertebrates are unable to actively search minutes of its arrival on the seabed and can reach for sources of fisheries-generated carrion and rely densities in excess of 300m-2 (Kaiser and Spencer on its chance occurrence on the seabed. Thus, 1996a; Ramsay et al. 1997). It is not surprising at present, this source of carrion may be too unpre- therefore that dominance hierarchies occur among dictable to give significant benefit to most inver- the scrum of scavenging invertebrates such that tebrate scavenger populations. In line with our smaller conspecifics are excluded from the food rationale for the relative lack of cascade effects resource as are other less competitive species observed in marine systems as a result of fish- (Ramsay et al. 1997; Kaiser et al. 1998). The ing activities, it is perhaps not surprising that we amount of carrion generated on the seabed by bot- only see population effects of subsidies at higher tom fishing is at least equivalent to the amount of trophic levels (sea birds) whereas they are not ap- material discarded at the surface of the sea parent, or are less obvious, at lower trophic levels. (Bergman and Van Santbrink 2000). However, this carrion is calculated to supply only 7% of the max- imum annual energy requirements of invertebrate 16.5 EFFECTS OF scavenging megafauna in the southern North Sea REMOVING PREDATORS (Fonds and Groenwold 2000). 16.5.1 The ‘krill surplus’ hypothesis While the responses of populations of scaveng- ing sea birds have been correlated in part with Commercial whaling dramatically reduced the subsidies from discarding practices, the responses abundance of baleen whales in the Antarctic of fish and invertebrate scavenger populations to Ocean, and these whales were the main consumers fisheries subsidies are far less clear (Ramsay et al. of krill (mostly Enphausia superba). The reduced 2000). Several studies indicate that populations predation on krill was expected to release krill pro- of flatfish species that are known to be facultative duction for species that formerly competed with scavengers have increased in abundance or in- whales. This was known as the ‘krill surplus hy- creased their growth rate in response to increased pothesis’. Predicted krill production in the south- fishing effort. For example, the growth rates of west Atlantic is 28.6 to 97.6 million tonnes yr-1, plaice (Pleuronectes platessa) and sole (Solea while sea bird, seal, whale and fish consumption in solea) have increased in relation to the increase in the same area was estimated as 32.6 million tonnes bottom fishing effort and hence disturbance of the yr-1 (Trathan et al. 1995). If the distributions and seabed (Millner and Whiting 1996; Rijnsdorp and feeding behaviour of krill feeders overlapped in Leeuwen 1996). However, there are many alterna- space and time then competition could reason- tive explanations for the increased growth of plaice ably be expected. However, a recent analysis by and sole, which include the reduction of the popu- Murphy (1995) suggests that many krill predators lations of their main predators and the enhance- cannot compete because they feed in different ment of polychaete communities due to the effects areas at different times. As a result, the effects of of eutrophication. competitive interactions between krill predators Why should there be such large differences in are minor in comparison with the oceanographic the responses of scavenging sea birds and fishes and factors that determine krill abundance on feeding invertebrates to energy subsidies? Sea birds are able grounds. In general, temporal and spatial variation to actively search for fishing vessels over large areas in krill abundance appears to drive the dynamics of and hence discards from trawlers are a relatively krill consumers. The variation is due to recruit- 354 Chapter 16 ment fluctuations and the differential transport of were calculated for the outcomes of interest, such krill by ocean currents. In the waters around South as changes in yields of different species following a Georgia, for example, krill abundance fluctuates seal cull. Contrary to the prediction of the simplis- by a factor of twenty between years, and in years of tic models, Yodzis showed that a cull of seals was low abundance krill predators suffer catastrophic more likely to be detrimental to total yields than breeding failures (Croxall et al. 1988; Brierley beneficial. et al. 1997; Murphy et al. 1998). The early models that predicted the existence of a ‘krill surplus’ were based on a homogeneous view of the food chain and assumed that krill were equally vulnerable 16.6 EFFECTS ON to predators when in fact they were not (Murphy BENTHIC COMMUNITIES 1995). AND HABITATS

16.5.2 Culling mammals to All continental shelf areas are subject to fishing boost fishery yields using towed bottom-fishing gear. Fishing in the marine environment is perhaps the best example Fishers see marine mammals eating fish and often of a method of harvesting animals that can lead treat them as competitors. Indeed, there are many to habitat change. In a recent article, Watling and examples of fishers culling marine mammals Norse (1998) compared the effects of fishing in an attempt to boost fishery yields (Earle 1996; with towed bottom-fishing gears to clear-cutting Hutchinson 1996). There are similar traditions in of forests in terrestrial systems. This is perhaps freshwater systems, where by anglers sometimes understandable, given the manner in which towed remove predatory birds and fishes in the hope that bottom fishing gears operate (see section 16.4 and this will increase their catches (Cowx, Chapter 17, Misund et al., Chapter 2, this volume). Watling this volume). and Norse’s (1998) assertion was that the inciden- Traditionally, the simple logic that marine tal or even deliberate removal of topographically mammals and fishers competed directly was complex seabed habitats would have detrimental followed, and the debate focused on how much effects on the associated species assemblage as in the mammals ate and whether this consumption terrestrial systems. This is perhaps not surprising was significant in relation to catch (Beddington and there is good evidence to demonstrate these et al. 1985; Crawford et al. 1992). This argument effects in some marine systems. For example, assumes a simple food web as perceived by many Sainsbury (1987; 1998) reported that as sponges and fishers, in which fishers compete with a marine soft-coral communities were removed as bycatch mammal for a prey. However, such systems are in trawls off the northwestern shelf of Australia, embedded in a much more complex system in the associated fish species were greatly reduced in which the interactions between fishers and marine abundance. Only after these areas were protected mammals do not follow the rules predicted by from bottom fishing did they observe slow regen- simple models (Beverton 1985; Yodzis 1998; Pauly eration of the habitat as sponges and corals began and Christensen, Chapter 10, this volume). to recolonize and grow. Once this process had A recent analysis by Yodzis (1998) considered begun, then the populations of associated fish the potential effects of a fur seal cull on fishery species began to increase once again. In temperate yields from the Benguela upwelling ecosystem off estuarine systems, oysters are important reef- West Africa. Here, the simple model, where seals forming organisms that add structural complexity and fishers compete for fish that feed on every- to the seabed and increase species and habitat thing else, was discarded, and Yodzis described the diversity. Oysters also improve water quality system using a food web containing 29 species or through their filtration activities as they bind species groups (Fig. 16.1). Probability distributions particulate organic matter and remove pollutants Ecosystem Effects of Fishing 355 and nutrients from the water column. The func- Using a combination of fishing effort data and tional role of oysters in coastal ecosystems is direct observations from side-scan sonar surveys, now deemed to be of such importance that in many Collie et al. (1997) were able to identify compar- areas large sums of money are spent attempting to able substrata that experienced different inten- re-establish degraded oyster reefs (Lenihan 1999). sities of scallop dredging on the Georges Bank, Many fisheries that use towed bottom gears northwest Atlantic. Areas that were less frequent- occur over coarse sediments in relatively shallow ly fished were characterized by abundant bryo- waters. The species that live in these habitats are zoans, hydroids and worm tubes which increased adapted to physical disturbance as they are regu- the three-dimensional complexity of the habitat. larly subjected to wave action, strong currents and Furthermore, examination of evenness within tidal scouring (Kaiser 1998). Hence, it might be the community suggested dominance by these expected that the effects of fishing disturbance in structural organisms, which indicated that this such habitats would be relatively minor in com- environment was relatively undisturbed. In con- parison with other more stable environments. The trast, the more intensively dredged areas had lower ecological significance of fishing disturbance will species diversity, lower biomass of fauna, and were vary according to the intensity, frequency and dominated by hard-shelled bivalves (e.g. Astarte history of disturbance that has occurred over a spp.), echinoderms and scavenging decapods. particular area of the seabed. This implies that in The higher diversity indices observed at the less some habitats a certain level of physical distur- intensively dredged sites were attributable to the bance may be sustainable, whereas in others it may large number of organisms, such as polychaetes, have long-lasting ecological effects. This is proba- shrimp, brittle stars, mussels and small fishes, that bly the most relevant question for current research were associated with the emergent sessile fauna as the implicit assumptions would form the (Collie et al. 1997). Many of these associated basis for predicting the likely outcome of bottom- species were also important prey for commercially fishing practices in different habitats. exploited fishes such as cod (Bowman and There now exists an extensive literature on Michaels 1984). This study emphasizes how fish- manipulative experiments designed to ascertain the ing activities can degrade habitat characteristics immediate effects of fishing disturbance on benthic on which commercial fish species depend for suc- community structure. The majority of thesestudies cessful growth and survival (see Benaka 1999). demonstrate significant short-term reductions in Bradshaw et al. (2000) were able to take advan- species number and counts of individual organisms tage of the creation of an area closed to fishing to as a result of fishing disturbance (Jennings and monitor subsequent changes in benthic commu- Kaiser 1998; Auster and Langton 1999). While nity structure. The closed area was located off the these studies have been important for understand- south of the Isle of Man, Irish Sea, in an intensively ing the mechanism of community change and as- fished scallop ground. After several years surveyed certaining those species that are most vulnerable plots within the closed area had become signifi- to disturbance, they are unable to replicate the in- cantly different to similar plots of seabed in the tensity, frequency or long-term disturbance histo- adjacent seabed that remained open to fishing. ry associated with the fishing activities of an entire Bradshaw et al. (2000) then tested the hypothesis fleet of fishing vessels. A better approach to study- that these changes had occurred as a result of the ing the long-term effects of fishing disturbance at lack of fishing disturbance. Within the area closed the scale of the fishing fleet is to compare areas sub- to fishing, experimental plots were fished with the jected to different known intensities of fishing dis- same scallop dredges that are used by the commer- turbance. A number of studies have employed this cial fleet. As a result of their experimental fishing, approach and have independently shown similar Bradshaw et al. (2000) altered the benthic commu- findings (Collie et al. 1997; Thrush et al. 1998; nity in the experimental plots within the closed Bradshaw et al. 2000; Kaiser et al. 2000a,b). area such that it became similar to that in the adja- 356 Chapter 16 cent commercial fishing ground. This manipula- bottom disturbances for 20 years. Standard benthic tion convincingly demonstrated that the seabed surveys of the infauna and epifauna within the surrounding the closed area was maintained in different areas revealed that the areas that had a permanently altered state by bottom-fishing been closed to towed bottom-fishing gears were disturbance. dominated by emergent sessile epifauna that had In a similar study, Kaiser et al. (2000a) investi- a relatively high biomass. In contrast, the areas of gated differences in benthic community structure seabed subjected to either seasonal or continuous and habitat complexity in areas exposed to differ- towed bottom-fishing activities were dominated ing levels of bottom-fishing activity according to by small-bodied opportunistic species and mobile the restrictions imposed by a voluntary manage- scavenging fauna (Fig. 16.5). ment agreement. This agreement exists between The resolution of fishing-effort data sets re- fixed-gear and towed bottom-gear fishers that op- mains a major obstacle to comparative studies of erate off the south Devon coast in England. The the long-term effects of repeated bottom-fishing agreement was instigated in 1978. All fishing by disturbance. The simplest solution is to install scallop dredgers and trawlers is prohibited in areas tracking devices to fishing boats that would record reserved for the setting of pots or fixed bottom gear. their precise position. These devices are now fitted Some areas are open to towed bottom gears for to a proportion of certain fleets for the purpose of limited periods of the year. The closed areas extend monitoring compliance with fisheries regulations out to 6 nautical miles from the shore, beyond (Rijnsdorp et al. 1998). However, these schemes which there are no operational restrictions on are in their infancy and do not yet provide a long- bottom-fishing activities. This management sys- term historical record of seabed disturbance. tem permitted Kaiser et al. (2000a) to compare Kaiser et al. (2000b) were able to partly overcome areas of the seabed that had been exposed to differ- some of these problems to compare areas of the ing levels of fishing disturbance over a prolonged seabed off the Isle of Man that are subjected to period. The areas open to pot fishing only were either high or low levels of scallop-dredging considered to have experienced only infrequent disturbance. They chose this fishery as the subject

(a) Low disturbance (b) Medium disturbance (c) High disturbance 100 100 100 90 90 90 80 B 80 80 70 70 70 60 A 60 60 50 50 50 40 40 40 30 30 30 20 20 20 Abundance/biomass (%) 10 10 10 0 0 0 1 10 100 1 10 1 10 100 Species rank

Fig. 16.5 Abundance/biomass curves of samples collected from (a) areas protected from towed bottom-fishing gear (low disturbance), (b) areas open seasonally to towed bottom-fishing gear and those areas (c) that are fished all year with towed bottom-fishing gear (high disturbance). As the level of bottom-fishing disturbance increases the biomass curve (B) converges with the abundance curve (A), which is a typical response in stressed communities. (Source: adapted from Kaiser et al. 2000a.) Ecosystem Effects of Fishing 357

Fig. 16.6 An acetate peel of a sec- tion through the shell of the bivalve Glycymeris glycymeris collected from a commercial scallop fishing ground. (A) Growth bands in the shell matrix are laid down each year, enabling the age of the bivalve to be determined; (B) physical damage from passing fishing gears is record- ed as a fracture in the outside portion of the shell which has then contin- ued to grow; (C) the inner side of the shell that is in direct contact with the bivalve’s tissues. (Source: adapted from Ramsay et al. 2000.)

of their investigation because fishing effort has experienced constantly high levels of fishing dis- been recorded at a relatively small spatial scale turbance were dominated by more opportunistic since the early 1980s (5 ¥ 5 nautical miles). Never- and scavenging species. theless, even this relatively fine scale of fishing- effort data presents problems of sampling at the 16.6.1 Meta-analysis: a basis for appropriate scale when one considers that the sam- better management pling devices used collect organisms from areas of 0.1m2 (grabs) to approximately 1m2 (dredges). As we have seen, the magnitude of the impact of Kaiser et al. (2000b) were able to corroborate the bottom-fishing activities on benthic fauna and disturbance history of the specific areas of seabed their habitat varies according to factors such as sampled from the disturbance scars that occurred habitat stability, depth, disturbance frequency and in the shells of a long-lived bivalve, Glycymeris presence of biogenic structures. Thus, the results glycymeris. Bivalves lay down annual growth from any one experiment or comparative study are rings, hence it was also possible to determine the limited in their scope for extrapolation to other sit- year in which the disturbance occurred. Fishing uations, and such extrapolations may be entirely gears that come into contact with bivalves leave misleading. A far better approach is to take all the distinctive scars in the shell matrix when the ani- data from these studies and undertake what is mal’s mantle withdraws in response to physical in- known as a meta-analysis of the data, which in- jury (Fig. 16.6). The occurrence of these scars in the creases the power for the analysis of general trends shells of G. glycymeris confirmed that the chosen and patterns that might provide the basis of predic- heavily fished areas had experienced significantly tive models (Collie et al. 2000; Myers Chapter 6, higher levels of fishing disturbance consistently Volume 1). Collie et al. (2000) undertook such an over many years compared with low mean levels of analysis and identified those gears that cause the disturbance experienced in the chosen areas of low greatest initial impact on benthic habitats. Note fishing effort. An examination of the benthic com- this does not equate to long-term impact. Not sur- munities in these areas showed that large biomass prisingly they found that intertidal dredging activ- emergent epifauna were more prevalent in the ities caused the greatest initial impacts, followed areas of seabed that had been fished sporadically by scallop dredging and then trawling. The most over many years, whereas those areas that had consistently interpretable result was with respect 358 Chapter 16

2 Mud2 Muddy sand

0 0

–2 –2

–4 –4 Fig. 16.7 Results from a meta- 1 5 10 50 500 1 5 10 50 500 analysis of the effects of fishing disturbance on benthic commu- 2 Sand2 Biogenic nities. The scatter plots of the relative change of all species (each datapoint represents the relative

Relative change in abundance 0 0 abundance of a different species on each different sampling date) in different habitats at time intervals –2 –2 after the occurrence of a fishing disturbance. The fitted curves show the predicted time trajectory forrecoverytooccur.On they-axis, –4 –4 0 shows no relative change in 1 5 10 50 500 1 5 10 50 500 abundance, negative values show Log days Log days a relative decrease in abundance. to faunal vulnerability, with a ranking of initial shows recovery in biogenic habitats to occur after impacts that matched expectations based on 500 days, there were few data on which to base morphology and behaviour. this analysis. We now know that recovery rates in More importantly, Collie et al. (2000) were also some biogenic habitats take from 5 to 10 years able to model recovery rate and speculated about (Sainsbury et al. 1998). the level of physical disturbance that might be sustainable in a particular habitat. Their results suggested that sandy sediment communities are 16.7 BYCATCHES AND able to recover within 100 days (Fig. 16.7), which GHOST-FISHING implies that they could perhaps withstand two to three incidents of physical disturbance per year In their comprehensive review of world bycatches, without changing markedly in character. This is Alverson et al. (1994) extracted data from over 800 the average predicted rate of disturbance for the published records. Although this represents the whole of, for example, the southern North Sea. most authoritative text written on this subject, it However, we know that effort is patchily distrib- is certainly by no means entirely comprehensive. uted and that some relatively small areas of the Bycatch and discard data are best documented for seabed are disturbed approximately eight times per nations that monitor fisheries and have sufficient year (Rijnsdorp et al. 1998). If these recovery rates resources to operate observer programmes on fish- estimates for sandy habitats are realistic then we ing vessels, which are beyond the means of poorer would predict that the communities in these heav- nations. Trawl fisheries accounted for most of the ily fished areas are held in a permanently altered records of bycatches (966) followed by drift-nets state. Although Collie et al.’s (2000) analysis and gill-nets (232), line fisheries (150), pot (83) and Ecosystem Effects of Fishing 359 purse seine (82) fisheries. An investigation of adult population was lost each year to a combina- which target species or target groups were associ- tion of natural and fishing mortality (Piatt and ated with most records of bycatches revealed that Nettleship, 1987). shrimp and groundfish species headed the list. In Albatrosses, and in particular the wandering particular, shrimp fisheries accounted for approxi- albatross, have been severely affected by the mately 35% of the global commercial fisheries activities of the southern bluefin tuna (Thunnus discards wasting approximately 27 million tonnes maccoyii) and Patagonian toothfish (Dissostichus per annum. eleginoides) longline fisheries. Birds are attracted to the baited lines as they are paid away from 16.7.1 Sea birds the vessel and dive after the baited hooks as they pass into the water. The Southern Ocean bluefin There are numerous reported incidents of sea birds tuna fleet deployed 107 million hooks annually captured and drowned in fishing gears (Dayton et and was estimated to have killed 44000 alba- al. 1995). In general, set-net, drift-net and longline trosses up to 1989 when conservation measures fisheries take the most birds (see also Misund et al., were introduced. Nevertheless, modelling studies Chapter 2, this volume). These fisheries often of wandering albatross populations predict that target large fish species such as adult tunas and unless fishing is reduced by more than 50%, salmonids that do not occur in the diet of sea birds. this species will soon be extinct (Weimerskirch Diving sea birds are particularly vulnerable to and Jouventin 1987; Brothers 1991). Currently, capture by drift- and set-nets and longlines. The there are only 13 breeding pairs of the Amsterdam nylon meshes of drift- and set-nets are designed to albatross, Diomedia amsterdamensis, and only be invisible underwater, hence birds have little 158 pairs of the short-tailed albatross, D. albatrus. chance of avoiding any nets set in the area where Both of these species are known to encounter they are feeding. longline fisheries and the latter have already The number of sea birds killed annually is quite been caught as incidental catch in Hawaiian staggering for some fisheries. It is not possible for waters. With so few remaining breeding adults us to list all the recorded instances here, but it in these populations the loss of even a few birds is sufficient to say that the examples given are pro- will jeopardize survival (Weimerskirch et al. bably repeated elsewhere around the world. For 1997). example, an estimated 210000–760000 sea birds were killed every year by Japanese drift-nets fish- 16.7.2 Marine reptiles ing in the North Pacific Ocean (Northridge 1991). Subsequently, the use of these nets was banned in Captures of sea turtles have been recorded in long- the early 1990s. Although such high incidental line, set-net and towed-gear fisheries. The survival catches are undesirable, they are not necessarily of some turtle species is also threatened by loss of unsustainable if they constitute a relatively small suitable habitat in which to lay their eggs through (<1%) proportion of the population. However, in coastal tourist developments, egg stealing and ille- some circumstances bird deaths associated with gal fishing for wild turtles. Turtles are particularly nets can constitute the greatest source of mortality vulnerable to entanglement and drowning in gill- for some local populations. For example, a small nets and associated ghost-fishing gear as the rough population of 1600 razorbills (Alca torda) breeds in skin on their head and flippers catches easily on Newfoundland, Canada and from 1981 to 1984 ap- the meshes of these nets (Carr 1987). Baited long- proximately 200 birds were killed in nets each lines set for swordfish in the Mediterranean take year. This was slightly more than the 10% of the approximately 20000 endangered species of tur- population that died each year through natural tles every year. Similarly shrimp trawl fisheries are causes. As a result, this population was considered a significant source of mortality for turtles (Poiner under threat of extinction as nearly a quarter of the et al. 1990). 360 Chapter 16

To date, relatively little information exists on tive with respect to the size class of animals cap- the biology or population dynamics of sea snakes. tured, gill-nets are associated with high numbers As a result, it is difficult to gauge whether the esti- of incidental captures of cetaceans. In the Sri mated 120000 sea snakes, comprising 10 different Lankan gill-net fishery, one dolphin is caught for species, caught in the prawn trawl fisheries in every 1.7 to 4.0 tonnes of tuna landed. This com- the Gulf of Carpentaria, Australia, represents a sig- pares very poorly with one dolphin for every 70 nificant threat to the survival of these species. tonnes landed in the eastern Pacific purse-seine Post-capture mortality varies from 10% to 40% fishery. Hence, while the use of gill-nets may have depending on species, and in common with sea tur- conservation benefits for target fish stocks, the tles, increases with tow duration (Ward 1996a,b; negative aspects of bycatches of larger organisms Brewer et al. 1998). may be counterproductive (Hall 1998).

16.7.3 Marine mammals 16.7.4 Ghost-fishing It is an inescapable fact that wherever fishers pur- Static fishing gear, such as pots and set-nets, are sue prey fish species they are likely to encounter usually deployed and then left unattended for peri- larger predatory organisms such as marine mam- ods of up to several days (Misund et al. Chapter 2, mals. As industrialization of fisheries has replaced this volume). During this time the gear is vulnera- less intensive forms of fishing, so incidental catch- ble to damage from bad weather or can be snagged es of species that are closely associated with the accidentally by passing vessels or even deliberate- target fish species has increased. Such was the case ly vandalized. Once fishing gear has been lost it in the eastern Pacific tuna purse-seine fishery that could continue fishing indefinitely. This phenom- began in the late 1950s (Misund et al. Chapter 2, enon is termed ‘ghost-fishing’. this volume). This fishery replaced the highly Just how long a gear is likely to continue ghost- species-specific pole and line fishery that had oper- fishing depends upon the circumstances of its loss. ated previously. It is thought that schools of tuna Gear that has been dragged off by trawlers or lost in follow herds of dolphin as they pursue smaller fish rough weather is likely to become tangled and will species, but the exact reason for the association is fish less effectively when it is eventually cut free not entirely understood (Hall 1998). In the early and allowed to sink back to the seabed. In other cir- days of this tuna fishery dolphin were inevitably cumstances, the gear may be held in an open fish- captured since their presence within the purse ing position when it becomes snagged on rocks or seine net ensured the capture of the tuna. Crude es- similar seabed features. In sheltered or deep-water timates of the number of dolphin killed during the situations, lost gear is likely to continue fishing for 1960s indicated fatalities of hundreds of thousands many years as it will not be subjected to damage by per year, which caused a population decline until wave action and abrasion (Carr et al. 1992). In con- the late 1970s. This sparked the ‘tuna–dolphin’ trast to the numerous records of incidental catches debate that culminated in the Marine Mammal in drift-nets and other set gears few incidents of Protection Act of 1972 passed by the US Congress. ‘ghost-fishing’ are reported or studied (Dayton et This act necessitated the adoption of good fish- al. 1995). The few studies that have been under- ing practices that involved ‘backing-down’, which taken reveal large differences in the behaviour of is a process to release dolphins from the net, and lost gear according to the circumstances and envi- the provision of independent observers aboard ronment in which the gear is lost. Kaiser et al. tuna vessels to record dolphin mortality (Hall (1996) reported that a net that was snagged on 1996). rocky substrata in the Irish Sea continued to catch Gill-nets are also used to catch tuna in inshore crustacea and fish for over a year. In the shallow areas. United Nations legislation has banned the clear waters off the Algarve, Portugal, Erzini et al. use of high-seas gill-nets. Although highly selec- (1997) observed that similar lost nets were over- Ecosystem Effects of Fishing 361 grown by algae, thus increasing their visibility and gathering momentum, and even in the brief period reducing their ability to catch fish. since our previous review (Jennings and Kaiser, Pots or traps tend to be constructed of robust 1998), there has been a book (Hall 1999) and four man-made materials and incorporate a rigid struc- major symposia (Dorsey and Pederson 1998; ture. This means that pots are likely to maintain Benaka 1999; Kaiser and de Groot 2000; Gislason their shape and hence capture efficiency for much et al. 2000) on this subject. The most rapid de- longer than lost nets. Animals that are trapped and velopments are now being made in understand- die within pots act as bait that attracts yet more ing the impacts of trawling disturbance on habitat scavengers into the lost pot. The ‘ghost-fishing’ po- and integrating ecosystem concerns into fishery tential of pots also varies according to their design. management practices. Habitat conservation is For example, Parrish and Kazama (1992) found that now a management objective in several fisheries. the majority of Hawaiian spiny lobster (Palinurus In the United States, for example, amendments to marginatus) and slipper lobster (Scyllarisdes the Magnuson–Stevens Fishery Conservation and squammosus) were adept at escaping traps. In Management Act require fisheries managers to de- contrast, Smolowitz (1978) studied mortalities of scribe and identify essential fish habitat, to mini- lobster (Homarus americanus) in parlour-type mize to the extent practicable adverse effects on traps that have non-return entrances to minimize such habitat caused by fishing and to identify other animal escape. They found that 12–25% of trapped actions to encourage the conservation and en- individuals died in these pots (Smolowitz 1978). hancement of such habitat (Kurland 1998). As the Little is known about the frequency of net or pot habitat requirements of a large number of species loss, which no doubt results from the reluctance of are described, and the potential impacts of gears fishers to report such incidents and the difficulty are assessed, it is likely that the use of towed fish- in undertaking quantitative surveys to estimate ing gears will be controlled on a much wider scale the prevalence of lost gear. Estimates of gear loss than ever before. would indicate that these are substantial in some fisheries. The fact that fishers have complained about the problem of ghost-fishing certainly REFERENCES indicates the scale of the problem. These com- plaints prompted a grapnel survey of the seabed on Alverson, D., Freeberg, M., Pope, J. and Murawski, S. Georges Bank, which yielded 341 actively fishing (1994) A Global assessment of Fisheries By-catch and Discards. Rome: FAO. ghost nets from 286 tows (Brothers 1992). The phe- Anker-Nilssen, T. (1992) Food Supply as a Determinant nomenon of ghost-fishing was clearly perceived of Reproduction and Population Development in to have negative effects on commercial stocks of Norwegian puffins Fratercula arctica. Trondheim: Greenland halibut (Reinhardtius hippoglossoides) University of Trondheim. by commercial fishers involved in this fishery. As a Anon. (1994) Report of the study group on seabird/fish in- result, they instigated their own voluntary clean- teractions. International Council for the Exploration of the Sea C.M. 1994/L: 3, 1–119. up programme (Bech 1995). The adverse effects of Auster, P.J. and Langton, R.W. (1999) The effects of lost pots are relatively easily averted by incorpo- fishing on fish habitat. In: L. Benaka (ed.) Fish rating escape panels into pot design and by using Habitat: Essential Fish Habitat and Rehabilitation. biodegradable materials. These technical mea- Bethesda, Md.: American Fisheries Society, pp. sures are cheap and simple to instigate and are 150–87. extremely effective. Bax, N.J. (1991) A comparison of the biomass flow to fish, fisheries and mammals in six marine ecosystems. ICES Marine Science Symposia 193, 217–24. Bech, G. (1995) Retrieval of lost gillnets at Ilulissat 16.8 CONCLUSIONS Kangia, NAFO, Scientific Council Research Docu- ment 95/6. The study of the ecosystem effects of fishing is Beddington, J.R., Beverton, R.J.H. and Lavigne, D.M. 362 Chapter 16

(1985) Marine Mammals and Fisheries. London: Final report to the European Commission study con- George Allen and Unwin. tract BIOECO/93/10, Netherlands Institute for Sea Benaka, L. (ed.) (1999) Fish habitat: Essential fish habitat Research, Texel. and rehabilitation. In: American Fisheries Society Camphuysen, C.J., Ensor, K., Furness, R.W., Garthe, S., Symposium, 22. Bethesda, MD: American Fisheries Huppop, O., Leaper, G., Offringa, H. and Tasker, M.L. Society. (1993) Seabirds Feeding on Discards in Winter in the Bergman, M.J.N. and Van Santbrink, J.W. (2000) Fishing North Sea. Netherlands Institute for Sea Research, mortality of populations of megafauna in sandy sedi- Den Burg, Texel. ments. In: M.J. Kaiser and S.J. de Groot (eds) Effects of Camphuysen, C.J. and Garthe, S. (1987) Distribution and Fishing on Non-target Species and Habitats: Biologi- scavenging habitats of Northern Fulmars in the North cal, Conservation and Socio-economic Issues. Oxford: Sea. ICES Journal of Marine Science 54, 654–83. Blackwell Science, pp. 49–68. Camphuysen, C.J. and Garthe, S. (1997) Distribution and Beverton, R.J.H. (1985) Analysis of marine mammal- scavenging habits of northern fulmars in the North fisheries interactions. In: J.R. Beverton, R.J.H. Sea. ICES Journal of Marine Science 54, 654–83. Beverton and D.M. Lavigne (eds) Marine Mammals and Camphuysen, C.J. and Garthe, S. (2000) Seabirds and Fisheries. London: George Allen and Unwin, pp. 3–33. commercial fisheries: population trends of piscivorous Bohnsack, J.A. (1982) Effects of piscivorous predator re- seabirds explained? In: M.J. Kaiser and S.J. de Groot moval on coral reef fish community structure. In: G.M. (eds) Effects of Fishing on Non-target Species and Caillet and C.A. Simenstad (eds) Gutshop ’81: Fish Habitats: Biological, Conservation and Socio- Food Habits and Studies. Seattle: University of economic Issues. Oxford: Blackwell Science, pp. 163–84. Washington, pp. 258–67. Carpenter, S.R., Kitchell, J.F., Hodgson, J.R., Cochran, Bowman, R.E. and Michaels, W.L. (1984) Food of seven- P.A., Elser, J.J., Elser, M.M., Lodge, D.M., Kretchmer, teen species of northwest Atlantic fish. Technical D., He X. and von Ende, C.N. (1987) Regulation of lake Memorandum: National Marine Fisheries Service – primary productivity by food web structure. Ecology North East Fisheries Council, 28. 68, 1863–76. Bradshaw, C., Veale, L.O., Hill, A.S. and Brand, A.R. Carr, A. (1987) Impact of nondegradable marine debris on (2000) The effects of scallop dredging on gravelly the ecology and survival outlook of sea turtles. Marine seabed communities. In: M.J. Kaiser and S.J. de Groot Pollution Bulletin 18, 352–6. (eds) The Effects of Fishing on Non-target Species and Carr, H.A., Blott, A.J. and Caruso, P.G. (1992) A study Habitats: Biological, Conservation and Socio- of ghost gillnets in the inshore waters of southern economic Issues. Oxford: Blackwell Science. New England. MTS ’92: Global Ocean Partnership. Brewer, D., Rawlinson, N., Eayrs, S. and Burridge, C. Washington, DC: Marine Technology Society, pp. 361–7. (1998) An assessment of bycatch reduction devices Carr, M.H. and Hixon, M.A. (1995) Predation effects on in a tropical Australian prawn trawl fishery. Fisheries early post-settlement survivorship of coral-reef fishes. Research 36, 195–215. Marine Ecology Progress Series 124, 31–42. Brierley, A.S., Watkins, J.L. and Murray, A.W.A. (1997) Cederholm, C.J., Kunze, M.D., Murota, T. and Sibatani, Interannual variability in krill abundance at South A. (1999) Pacific salmon carcasses: essential contribu- Georgia. Marine Ecology Progress Series 150, 87–98. tions of nutrients and energy for aquatic and terrestrial Brothers, G. (1992) Lost or Abandonned Fishing Gear in ecosystems. Fisheries 24, 6–15. the Newfoundland Aquatic Environment. St Johns, Collie, J.S., Escanero, G.A. and Valentine, P.C. (1997) Newfoundland: Department of Fisheries and Oceans. Effects of bottom fishing on the benthic megafauna Brothers, N. (1991) Albatross mortality and associated of Georges Bank. Marine Ecology Progress Series 155, bait loss in Japanese longline fishery in the Southern 159–72. Ocean. Biological Conservation 55, 255–68. Collie, J.S., Hall, S,J., Kaiser, M.J. and Poiner, I.R. (2000) A Caley, M.J. (1993) Predation, recruitment and the dynam- quantitative analysis of fishing impacts on shelf-sea ics of communities of coral-reef fishes. Marine Biology benthos. Journal of Animal Ecology 69, 785–98. 117, 33–43. Crawford, R.J.M., Underhill, L.G., Raubenheimer, C.M., Camphuysen, C.J. (1993) Foerageermogelijkheden voor Dyer, B.M. and Martin, J. (1992) Top predators in the zeevogels in de boomkorvisserij; een verkennend Benguela ecosystem-implications of their trophic onderzoek. Sula 7, 81–104. position. South African Journal of Marine Science 12, Camphuysen, C.J., Calvo, B., Durninck, J., Ensor, K., 675–87. Follestad, A., Furness, R.W., Garthe, S., Leaper, G., Croxall, J.P., McCann, T.S., Prince, P.A. and Rothery, P. Skov, H., Tasker, M.L. and Winter, C.J.N. (1995) Con- (1988) Reproductive performance of seabirds and seals sumption of Discards by Seabirds in the North Sea. at South Georgia and Signy Island, South Orkney Ecosystem Effects of Fishing 363

Islands, 1976–1987: implications for Southern Ocean Predators and their Prey. Oxford: Blackwell Scientific monitoring studies. In: D. Sahrhage (ed.) Antarctic Publications, pp. 168–73. Ocean and Resources Variability. Berlin: Springer- Furness, R.W. and Monaghan, P. (1987) Seabird Ecology. Verlag, pp. 261–85. Glasgow: Blackie. Cruetzberg, F., Duineveld, G.C.A. and van Noort, G.J. Gislason, H., Sinclair, M.M., Sainsbury, K. and O’Boyle, (1987) The effect of different numbers of tickler chains R. (2000) Symposium overview: incorporating eco- on beam trawl catches. Journal du Conseil Inter- system objectives within fisheries management. ICES national pour l’Exploration de la Mer 43, 159–68. Journal of Marine Science 57, 468–75. Daan, N., Bromley, P.J., Hislop, J.R.G. and Nielsen, N.A. Greenstreet, S.P.R. and Hall, S.J. (1996) Fishing and (1990) Ecology of North Sea fish. Netherlands Journal ground-fish assemblage structure in the north-western of Sea Research 26, 343–86. North Sea: an analysis of long-term and spatial trends. Davis, G., Haaker, P. and Richards, D. (1996) Status and Journal of Animal Ecology 65, 577–98. trends of white abalone at the California Channel Is- Hall, M.A. (1996) On bycatches. Reviews in Fish Biology lands. Transactions of the American Fisheries Society and Fisheries 6, 319–52. 125, 42–8. Hall, M.A. (1998) An ecological view of the tuna–dolphin Dayton, P.K. and Hessler, R.R. (1972) Role of biological problem: impacts and trade-offs. Reviews in Fish disturbance in maintaining diversity in the deep sea. Biology and Fisheries 8, 1–34. Deep-Sea Research 19, 199–208. Hall, S.J. (1999) The Effects of Fishing on Marine Dayton, P.K., Thrush, S.F., Agardy, M.T. and Hofman, Ecosystems and Communities. Oxford: Blackwell R.J. (1995) Environmental effects of marine fishing. Science. Aquatic Conservation 5, 205–32. Hamer, K.C., Furness, R.W. and Caldow, R.G.W. (1991) Dorsey, E.M. and Pederson, J. (eds) (1998) Effects of Fish- The effects of changes in food availability on the ing Gear on the Sea Floor of New England. Boston, MA: breeding ecology of great skuas Catharacta skua Conservation Law Foundation. in Shetland. Journal of Zoology, London 223, 175– Earle, M. (1996) Ecological interactions between 88. cetaceans and fisheries. In: M.P. Simmonds and J.D. Hiatt, R.W. and Strasburg, D.W. (1960) Ecological rela- Hutchinson (eds) The Conservation of Whales and tionships of the fish fauna on coral reefs of the Marshall Dolphins. Chichester: John Wiley, pp. 167–204. Islands. Ecological Monographs 30, 65–127. Erzini, K., Monteiro, C., Ribeiro, J., Santos, M., Gaspar, Hill, B. and Wassenberg, T. (1990) Fate of discards from M., Monteiro, P. and Borges, T. (1997) An experimental prawn trawlers in Torres Strait. Australian Journal of study of gill net and trammel net ‘ghost-fishing’ off the Marine and Freshwater Research 41, 53–64. Algarve (southern Portugal). Marine Ecology Progress Hixon, M. (1991) Predation as a process structuring coral Series 158, 257–65. reef fish communities. In: P. Sale (ed.) The Ecology FAO (1994) Review of the state of world marine fishery of Fishes on Coral Reefs. San Diego: Academic Press, resources, Rome: FAO, pp. 1–136. pp. 475–508. Flecker, A.S. and Townsend, C.R. (1994) Community Hixon, M.A. and Beets, J.P. (1993) Predation, prey refuges wide consequences of trout introduction in New and the structure of coral reef fish assembleges. Ecolog- Zealand streams. Ecological Applications 4, 798–807. ical Monographs 63, 77–101. Fonds, M. and Groenwold, S. (2000) Food subsidies gener- Holland, K.N., Peterson, J.D., Lowe, C.G. and Wetherbee, ated by the beam-trawl fishery in the southern North B.M. (1993) Movements, distribution and growth rates Sea. In: M.J. Kaiser and S.J. de Groot (eds) Effects of of the white goatfish Mulloides flavolineatus in a fish- Fishing on Non-target Species and Habitats: Biologi- eries conservation zone. Bulletin of Marine Science 52, cal, Conservation and Socio-economic Issues. Oxford: 982–92. Blackwell Science, pp. 130–50. Hudson, A.V. and Furness, R.W. (1988) Utilization of dis- Furness, R. (1990) A preliminary assessment of the quan- carded fish by scavenging seabirds behind whitefish tities of Shetland sandeels taken by seabirds, seals, trawlers in Shetland. Journal of Zoology, London 215, predatory fish and the industrial fishery in 1981–83. 151–66. Ibis 132, 205–17. Hutchinson, J. (1996) Fisheries interactions: the harbour Furness, R., Ensor, K. and Hudson, A. (1992) The use of porpoise – a review. In: M.P. Simmonds and J.D. fishery waste by gull populations around the British Hutchinson (eds) The Conservation of Whales and Isles. Ardea 80, 105–13. Dolphins. Chichester: John Wiley, pp. 128–65. Furness, R.W. (1996) A review of seabird responses to Jennings, S. and Kaiser, M. (1998) The effects of fishing on natural or fisheries-induced changes in food supply. marine ecosystems. Advances in Marine Biology 34, In: S.P.R. Greenstreet and M.L. Tasker (eds) Aquatic 201–352. 364 Chapter 16

Jennings, S. and Polunin, N.V.C. (1996) Impacts of fishing salmon: II.15N and 13C evidence in the Kvichak River on tropical reef econsystems. Ambio 25, 44–9. watershed, Bristol Bay, south-western Alaska. Canadi- Jennings, S. and Polunin, N. (1997) Impacts of predator an Journal of Fisheries and Aquatic Science 50, depletion by fishing on the biomass and diversity of 2350–65. non-target reef fish communities. Coral Reefs 16, Kurland, J.M. (1998) Implications of the essential fish 71–82. habitat provisions of the Magnuson–Stevens Act. In: Jennings, S., Reynolds, J. and Mills, S. (1998) Life history E.M. Dorsey and J. Pederson (eds) Effects of Fishing correlates of responses to fisheries exploitation. Pro- Gear on the Sea Floor of New England. Boston, MA: ceedings of the Royal Society Series B 265, 333–9. Conservation Law Foundation, pp. 104–6. Jones, G.P., Millicich, M.J., Emslie, M.J. and Lunow, C. Lenihan, H.S. (1999) Physical-biological coupling on (1999) Self-recruitment in a coral reef fish population. oyster reefs: How habitat structure influences individ- Nature 402, 802–4. ual performance. Ecological Monographs 69, 251–75. Kaiser, M.J. (1998) Significance of bottom-fishing distur- Lloyd, C.S., Tasker, M.L. and Partridge, K. (1991) The bance. Conservation Biology 12, 1230–5. status of seabirds in Britain and Ireland. London: T. Kaiser, M.J., Bullimore, B, Newman, P, Lock, K, Gilbert, S and A.D. Poyser. (1996) Catches in ‘ghost-fishing’ set nets. Marine Ecol- Marten, G.G. (1979) Predator removal: effects on fish- ogy Progress Series 145, 11–16. eries yields in Lake Victoria (East Africa). Science 203, Kaiser, M.J. and De Groot, S.J. (2000) Effects of Fishing on 646–8. Non-target Species and Habitats: Biological, Conser- McClanahan, T.R. (1992) Resource utilization, competi- vation and Socio-economic Issues. Oxford: Blackwell tion and predation: a model and example from coral Science. reef grazers. Ecological Modelling 61, 195–215. Kaiser, M.J. and Ramsay, K. (1997) Opportunistic feeding McClanahan, T. (1995a) A coral-reef ecosystem-fisheries by dabs within areas of trawl disturbance: possible model – impacts of fishing intensity and catch selec- implications for increased survival. Marine Ecology tion on reef structure and processes. Ecological Model- Progress Series 152, 307–10. ling 80, 1–19. Kaiser, M.J., Ramsay, K., Richardson, C.A., Spence, F.E. McClanahan, T. (1995b) Fish predators and scavengers of and Brand, A.R. (2000b) Chronic fishing disturbance the sea urchin Echinometra mathaei in Kenyan coral- has changed shelf sea benthic community structure. reef marine parks. Environmental Biology of Fishes 43, Journal of Animal Ecology 69, 494–503. 187–93. Kaiser, M.J. and Spencer, B.E. (1994) Fish scavenging McClanahan, T., Kakamura, A., Muthiga, N., Yebio, M. behaviour in recently trawled areas. Marine Ecology and Obura, D. (1996) Effects of sea-urchin reductions Progress Series 112, 41–9. on algae, coral and fish populations. Conservation Kaiser, M.J. and Spencer, B.E. (1995) Survival of by-catch Biology 10, 136–54. from a beam trawl. Marine Ecology Progress Series 126, McClanahan, T.R. (1988) Coexistence in a sea urchin 31–8. guild and its implications to coral reef diversity and Kaiser, M.J. and Spencer, B.E. (1996a) The effects of beam- degradation. Oecologia 77, 210–18. trawl disturbance on infanunal communities in differ- McClanahan, T.R. and Muthiga, N.A. (1988) Changes in ent habitats. Journal of Animal Ecology 65, 348–58. Kenyan coral reef community structure and function Kaiser, M.J. and Spencer, B.E. (1996b) The behavioural due to exploitation. Hydrobiologia 166, 269–76. response of scavengers to beam-trawl disturbance. In: McClelland, B.R., Young, L.S., Shea, D.S., McClelland, S. Greenstreet and M. Tasker (eds) Aquatic Pre- P.T., Allen, H.L. and Spettigue, E.B. (1982) The bald dators and their Prey. Oxford: Blackwell Scientific eagle concentration in Glacier National Park, Publications. Montana: origin, growth and variation in numbers. Kaiser, M.J., Spence, F.E. and Hart, P.J.B. (2000a) Fishing The Living Bird 19, 133–55. gear restrictions and conservation of benthic habitat McIntosh, A.R. and Townsend, C.R. (1996) Interactions complexity. Conservation Biology 14, 1512–25. between fish, grazing invertebrates and algae in a New Kirkwood, G., Beddington, J. and Rossouw, J. (1994) Zealand stream: a trophic cascade mediated by fish Harvesting species of different lifespans. In: P. Edwards, induced changes in behaviour. Oecologia 108, 174–81. R. May and N. Webb (eds) Large-scale Ecology and Millner, R. and Whiting, C. (1996) Long-term changes in Conservation Biology. Oxford: Blackwell Scientific, growth and population abundance of sole in the North pp. 199–227. Sea from 1940 to the present. ICES Journal of Marine Kline, T.C., Goering, J.J., Mathisen, O.A., Poe, P.H., Science 53, 1185–95. Parker, P.L. and Scalan, R.S. (1993) Recycling of Monaghan, P. (1992) Seabirds and sandeels – the conflict elements transported upstream by runs of Pacific between exploitation and conservation in the northern Ecosystem Effects of Fishing 365

North Sea. Biodiversity and Conservation 1, 98– Pauly, D., Christensen, V., Dalsgaard, J., Froese, R. and 111. Torres, F. (1998) Fishing down marine food webs. Monaghan, P., Uttley, J. and Burns, M. (1992) The effects Science 279, 860–3. of changes in food availability on reproductive effort in Piatt, J.F. and Nettleship, D.N. (1987) Incidental catch arctic terns Sterna paradisaea. Ardea 80, 71–81. of marine birds and mammals in fishing nets off Murphy, E.J. (1995) Spatial structure of the Southern Newfoundland Canada. Marine Pollution Bulletin Ocean ecosystem: predator–prey linkages in Southern 18, 344–9. Ocean food webs. Journal of Animal Ecology 64, Poiner, I.R., Buckworth, R. and Harris, A. (1990) Inciden- 333–47. tal capture and mortality of sea turtles in Australia’s Murphy, E., Springer, A. and Roseneau, D. (1991) northern prawn fishery. Australian Journal of Marine High annual variability in reproductive success of and Freshwater Research 41, 97–110. kittywakes (Rissa tridactyla L.) at a colony in Polis, G. and Strong, G. (1996) Food web complexity and western Australia. Journal of Animal Ecology 60, community dynamics. The American Naturalist 147, 515–34. 813–46. Murphy, E.J., Watkins, J.L., Reid, K., Trathan, P.N., Ramsay, K., Kaiser, M.J. and Hughes, R.N. (1996) Everson, I., Croxall, J.P., Priddle, J., Brandon, M.A., Changes in hermit crab feeding patterns in response to Brierley, A.S. and Hofmann, E. (1998) Interannual vari- trawling disturbance. Marine Ecology Progress Series ability of the South Georgia marine ecosystem: biolog- 144, 63–72. ical and physical sources of variation in the abundance Ramsay, K., Kaiser, M.J., Moore, P.G. and Hughes, R.N. of krill. Fisheries Oceanography 7, 381–90. (1997) Consumption of fisheries discards by benthic Myers, R., Hutchings, J. and Barrowman, N. (1996) scavengers: utilisation of energy subsidies in different Hypothesis for the decline of cod in the North marine habitats. Journal of Animal Ecology, 86, Atlantic. Marine Ecology Progress Series 138, 293– 884–96. 308. Ramsay, K., Kaiser, M.J., Rijnsdorp, A.D., Craeymeersch, Nehlsen, W., Williams, J.E. and Lichatowich, J.A. (1991) J. and Ellis, J. (2000) Impact of trawling on populations Pacific salmon at the crossroads: stocks at risk from of the invertebrate scavenger Asterias rubens. In: M.J. California, Oregon, Idaho, and Washington. Fisheries Kaiser and S.J. de Groot (eds) Effects of Fishing on Non- 16, 4–21. target Species and Habitats: Biological, Conservation Northridge, S. (1991) Driftnet fisheries and their impacts and Socio-economic Issues. Oxford: Blackwell Sci- on non-target species: a worldwide review. Food and ence, pp. 151–62. Agriculture Organization of the United Nations, Vol. Ramsay, K., Kaiser, M.J., Richardson, C.A., Veale, L.O. 320, pp. 1–115. and Brand, A.R. (2000) Can shell scars on dog cockles Pace, M.L., Cole, J.J., Carpenter, S.R. and Kitchell, (Glycymeris glycymeris L.) be used as an indicator of J.F. (1999) Trophic cascades revealed in diverse fishing disturbance? Journal of Sea Research 43, ecosystems. Trends in Ecology and Evolution 14, 167–76. 483–8. Rice, J. and Gislason, H. (1996) Patterns of change in the Parrish, F.A. and Kazama, T.K. (1992) Evaluation of ghost size spectra of numbers and diversity of the North Sea fishing in the Hawaiian lobster fishery. Fisheries Bul- fish assemblage, as reflected in surveys and models. letin 90, 720–5. ICES Journal of Marine Science 53, 1214–25. Parrish, J., Norris, J., Callahan, M., Magarifugi, E. and Rijnsdorp, A.D., Buijs, A.M., Storbeck, F. and Visser, E. Schroeder, R. (1986) Piscivory in a coral reef commun- (1998) Micro-scale distribution of beam trawl effort in ity. In: G. Caillet and C. Simenstad (eds) Gutshop ’81: the sourthern North Sea between 1993 and 1996 in re- Fish Food Habits and Studies. Seattle: University of lation to the trawling frequency of the sea bed and the Washington, pp. 73–8. impact on benthic organisms. ICES Journal of Marine Parrish, J.D., Callahan, M.W. and Norris, J.E. (1985) Fish Science 55, 403–19. trophic relationships that structure reef communities. Rijnsdorp, A.D. and Leeuwen, P.I. Van (1996) Changes Proceedings of the Fifth International Coral Reef in growth of North Sea plaice since 1950 in relation Symposium 4, 73–8. to density, eutrophication, beam-trawl effort, and Pascoe, S. (2000) Economic incentives to discard by- temperature. ICES Journal of Marine Science 53, catch in unregulated and individual transferable 1199–213. quotas fisheries. In: M.J. Kaiser and S.J. de Groot (eds) Russ, G.R. (1985) Effects of protective management on Effects of Fishing on Non-target Species and Habitats: coral reef fishes in the central Philippines. Proceedings Biological, Conservation and Socio-economic Issues. of the Fifth International Coral Reef Symposium 4, Oxford: Blackwell Science, pp. 315–31. 219–24. 366 Chapter 16

Russ, G.R. (1991) Coral reef fisheries: effects and yields. M.W. (1995) An Atlas of Seabird Distribution in In: P.F. Sale (ed.) The Ecology of Fishes on Coral Reefs. North-west European Waters. Peterborough: Joint San Diego: Academic Press, pp. 601–35. Nature Conservation Committee. Russ, G.R. and Alcala, A.C. (1998a) Natural fishing ex- Swearer, S.E., Caselle, J.E., Lea, D.W. and Warner, periments in marine reserves 1983–1993: community R.R. (1999) Larval retention and recruitment in an and trophic responses. Coral Reefs 17, 383–97. island population of a coral-reef fish. Nature 402, Russ, G.R. and Alcala, A.C. (1998b) Natural fishing 799–802. experiments in marine reserves 1983–1993: roles of Tasker, M.L., Webb, A., Hall, A.J., Pienkowski, M.W. life history and fishing intensity in family responses. and Langslow, D.R. (1987) Seabirds in the North Sea. Coral Reefs 17, 399–416. Peterborough: Nature Conservation Council UK. Sainsbury, K.J. (1987) Assessment and management of Thrush, S.F., Hewitt, J.E., Cummings, V.J., Dayton, P.K., the demersal fishery on the continental shelf of north- Cryer, M., Turner, S.J., Furnell, G., Budd, R., Milburn, western Australia. In: J.J. Polovina and S. Ralston (eds) C. and Wilkinson, M.R. (1998) Disturbance of the ma- Tropical Snappers and Groupers – Biology and Fish- rine benthic habitat by commercial fishing: impacts at eries Management. Boulder, Col.: Westview Press, the scale of the fishery. Ecological Applications 8, pp. 465–503. 866–79. Sainsbury, K.J., Campbell, R.A., Lindholm, R. and Van Blaricom, G.R. (1982) Experimental analysis of Whitlaw, A.W. (1998) Experimental management of structural regulation in a marine sand community an Australian multispecies fishery: examining the pos- exposed to oceanic swell. Ecological Monographs sibility of trawl-induced habitat modification. In: E.K. 52, 283–305. Pikitch, D.D. Huppert and M.P. Sissenwine (eds) Ward, T.M. (1996a) Sea snake by-catch of prawn trawlers Global Trends: Fisheries Management. Bethesda, Md: on the northern Australian continental shelf. Marine American Fisheries Society, pp. 107–12. and Freshwater Research 47, 625–30. Samoilys, M. (1988) Abundance and species richness of Ward, T.M. (1996b) Sea snake by-catch of prawn trawlers coral reef fish on the Kenyan coast: the effects of pro- on the northern Australian continental shelf. Marine tective management and fishing. Proceedings of the and Freshwater Research 47, 631–5. Sixth International Coral Reef Symposium 2, 261–6. Watling, L. and Norse, E.A. (1998) Disturbance of the Sherman, K., Jones, C., Sullivan, L., Smith, W., Berrien, P. seabed by mobile fishing gear: a comparison to forest and Ejsymont, L. (1981) Congruent shifts in sandeel clearcutting. Conservation Biology 12, 1180–97. abundance in western and eastern North Atlantic Weimerskirch, H., Brothers, N. and Jouventin, P. (1997) ecosystems. Nature 291, 487–9. Population dynamics of wandering albatross Smolowitz, R.J. (1978) Trap design and ghost fishing: An Diomedea exulans and Amsterdam albatross D. overview. Marine Fisheries Review 40, 2–8. amsterdamensis in the Indian Ocean and their rela- Sprules, W. and Goyke, A. (1994) Size-based structure tionships with long-line fisheries: conservation and production in the pelagia of Lakes Ontario and implications. Biological Conservation 79, 257–70. Michigan. Canadian Journal of Fisheries and Weimerskirch, H. and Jouventin, P. (1987) Population dy- Aquatic Science 51, 2603–11. namics of the wandering albatross, Diomedia exulans, Sprules, W. and Stockwell, J. (1995) Size based biomass of the Crozet Islands: causes and consequences of the and production models in the St Lawrence Great population decline. Oikos 49, 195–213. Lakes. ICES Journal of Marine Science 52, 705–10. Willson, M.F. and Halupka, K.C. (1995) Anadromous fish Sprules, W.G., Casselman, J.M. and Shuter, B.J. (1983) as keystone species in vertebrate communities. Con- Size distributions of pelagic particles in lakes. Canadi- servation Biology 9, 489–97. an Journal of Fisheries and Aquatic Sciences 40, Yodzis, P. (1998) Local trophodynamics and the interac- 1761–5. tion of marine mammals and fisheries in the Benguela Stockton, W.L. and DeLaca, T.E. (1982) Food falls in the ecosystem. Journal of Animal Ecology 67, 635–58. deep sea: occurrence, quality and significance. Deep- Zeller, D.C. and Russ, G.R. (1998) Marine reserves: Sea Research 29, 157–69. patterns of adult movement of the coral trout (Plec- Stone, C.J., Webb, A., Barton, C., Ratcliffe, N., Reed, T.C., tropomus leopardus) (Serranidae). Canadian Journal Tasker, M.L., Camphuysen, C.J. and Pienkowski, of Fisheries and Aquatic Science 55, 917–24. 17 Recreational Fishing

I.G. COWX

17.1 INTRODUCTION be kept away from those who were not gentlemen and who, through immoderation in angling, might Definitions of recreational fishing vary depending ‘utterly destroye it’. on the origins and diversity of the activity. The The importance of recreational fishing as a FAO (1997), for the purposes of reporting catch sta- sport or leisure activity emanates from the 16th tistics, defined recreational fisheries as: ‘Fisheries and 17th centuries, and coincides with the publi- conducted by individuals primarily for sport but cation of Izaak Walton’s The Compleat Angler, or with a possible secondary objective of capturing Contemplative Man’s Recreation in 1653. This fish for domestic consumption but not for onward volume fixed for all time the position of the angler sale.’ Thus recreational fisheries involve both sub- who loves fishing for the sake of fishing; a position sistence fishing, where the catch is consumed, and summarized by Carlton (1975) as: ‘recreational leisure fishing, where the fish are returned live to fishing is the ritual pursuit of pleasure associated the water. Both types of recreational fishing are with the experience’. This definition has now been not only extremely important activities, but also superseded, with a change in demand for leisure are valuable resources contributing significantly activities. The sport is now highly developed and to national economies. Pitcher (1999), however, pursued by large numbers of people around the considered that recreational fishing should essen- world, primarily for pleasure, but also for income tially be for ‘fun’, based on the premise that ‘it costs generation and to supplement food supply, as ex- far more to catch the fish than they can possibly be emplified by the following statistics. worth as marketed commodities’. Against this ar- • Amongst 22 European countries there are an gument it seems perverse that anglers often endure estimated 21.3 million anglers, with an estimated torturous conditions to pursue their sport! expenditure on recreational fishing in 10 of the The true origins of recreational fishing (i.e. an- countries in Western Europe where data were gling, like hunting) began as a means of obtaining available, in excess of US$10 billion (Cowx 1998b). food. It was not until 1496 that angling as a true • In 1996, 18% of the US population 16 years of age sport or recreation can be interpreted from the and older, i.e. 35 million persons, exerted 514 literature. In that year, the Treatyse of Fysshynge million angler-days in fresh waters, expending wyth an Angle, attributed to Dame Juliana US$38.0 billion (US Fish and Wildlife Service Berners, a sportswoman and nun, was published 1997). in the second Book of St. Albans, a treatise on • In Canada in 1995, 4.2 million anglers exerted hawking, hunting and heraldry. These were major 55.5 million days and caught over 254 million interests of nobility and the publisher, Wynkyn de fishes while spending US$5.1 billion of which Worde, emphasized that the book was intended to US$3.4 billion was directly associated with the 368 Chapter 17 sport. Of these fishes some 113 million were 17.2.1 Fishing gears retained (Department of Fisheries and Oceans, Canada, 1998). Most recreational fishing is carried out using rod • It is estimated that total recreational catch and line, or a variation on the theme (Wilson 1999). worldwide is of the order of 2 million tonnes, and In the past 30 years, the gears used by recreational represents an important source of animal protein fishers have developed out of all recognition, from in many developing countries (Coates 1995). traditional cane and glass fibre rods, towards the This chapter examines the status of recreation- use of modern materials, such as carbon fibre, and al fishing around the world, mechanisms for as- technology such as echosounding to detect and sessing the status of recreational fisheries, current catch fish. The line used nowadays is almost exclu- management methods for recreational fisheries, sively monofilament nylon, and hooks are baited some of the issues and problems facing recrea- with a diverse range of foods, both natural and tional fisheries, and future needs for maintaining artificial. The natural baits include maggots, meal the status of the subsector. paste, fish, worms or shrimp, although the more sophisticated anglers use a concoction of ingredi- ents to improve the chances of catching fish. They 17.2 TYPES OF may also use artificial flies and lures, particularly RECREATIONAL FISHING in game fisheries for salmon and trout (Salmo and Oncorhynchus spp.) or sailfish and marlin. These Exploitation of living aquatic resources for recrea- materials and methods have allowed the practi- tional and leisure purposes has a long tradition in tioner to increase the possibility of encountering temperate countries. However, recreational fish- and capturing fish, and to some extent have re- ing is now becoming common in some tropical moved some of the skill of the sport (Wilson 1999). countries. Brazil, the Great Lakes of Africa and In fresh waters, fishing is carried out from the southeast Asia, for instance, have witnessed huge bank, from small boats or wading in rivers or diversification of interest towards recreational shallow lakes. Sea fishing is carried out both from fishing to satisfy the needs of the growing urban the shore, on beaches and off rocky coastline, and populations, but also as a result of expansion in from inshore and offshore boats, depending on the ecotourism linked to diversification of sport fish- species being targeted. ing opportunities. Many people are now targeting Recreational fishing in some countries (e.g. the major rivers and lakes of the world for large- Canada, Finland, Sweden, Eastern Europe) is prac- sized fish (e.g. Nile perch, Lates niloticus, in Lake tised with gears that were predominantly designed Victoria), or those species that put up a good fight, for commercial purposes, such as gill-nets and e.g. mahseer (Tor tor) species in the Himalayan traps. Although classified as recreational fishing rivers of the Indian subcontinent or tiger fish (Hy- by the host countries, there is some debate as to the drocynus vittatus) in East African lakes. Similarly, validity of the claims because considerable har- interest in coastal game fisheries targeting marlin vests of fish are removed in these fisheries, (Makaira spp.), sailfish (Istiophorus spp.), and and in some cases they have contributed towards other large species has exploded with the growing the demise of the fish stocks (e.g. Müller 1990; affluence of the 20th century (e.g. McGrath et al. Gautman and Hicks 1999). Furthermore, regula- 1997). Recreational fishing, however, comprises a tions over the fishing activities have more in diverse array of approaches, which can be broken common with commercial than recreational down into exploitation of natural and enhanced fishing practices. Access and catch restrictions, are fisheries for pleasure, competition or supple- being used to regulate the salmonid sport fisheries menting domestic food supply, both in the marine of Canada (Walters and Cox 1999), which suggest and freshwater environments. that this is the case. Recreational Fishing 369

17.2.2 Recreational fishing classified ered a bonus from the experience. In recent years, by objective however, there has been a trend towards wanting to catch large numbers of fish. This is coupled with the According to the FAO (1997) definition, partici- decline in many natural fisheries, and movement pants in recreational fishing do not depend directly away from classical leisure fishing venues around on the fishery for employment, treating fishing rivers and coasts to intensively stocked fisheries more as a pastime. Exponents come from all levels where big catches are guaranteed (North 2002). of society, from juveniles through to the old age groups, but the type of fishing they participate in Match fishing often reflects their social status. Game fishing, either in fresh or marine waters, for example, is Match fishing is the competitive pursuit of angling usually dominated by the wealthier sectors of to catch the biggest weight of fish or the largest society coming from professional backgrounds fish in a defined period of time. Fishermen either because of the high costs of participation. This is perform individually or form groups that com- particularly true of the salmon fisheries of north- pete according to a specified set of rules, usually ern Europe and the offshore marlin and sailfish defining the gear and bait to be used, and the fisheries throughout the world. Conversely, fresh- species to be caught. Such fishing, although water fishing for cyprinids and fishing off the shore mainly practised in Europe and North America, is was, and still is, a major participatory sport of the highly organized and culminates in international middle and working classes. Recreational fishers competitions and world championships. A small can be categorized according to the way they ex- number of highly successful participants are able ploit the main target fish species, although care to earn considerable sums of money by winning must be taken not to compartmentalize persons, competitions, or make a living by endorsing an- because they often pursue more than one type of gling products, fishing venues or writing popular recreational fishing activity. articles for the media. However, the majority com- pete for the prestige gained through winning. Match fishing takes place in all types of waters, Leisure fishing both natural and stocked. Competitors are usually Angling for leisure remains one of the primary assigned a fixed fishing station or peg when based reasons for participating in the activity. Numerous from the shore or bank, but when fishing off a boat studies confirm this point and consistently cite this can be randomly positioned to pursue the best that a high percentage of fishermen go fishing to be catch. The premise underwriting match competi- able to relax in pleasant surroundings and are not tions from fixed stations is that there is potentially necessarily bothered as to the quantity of their an equity of catches between stations, thus the catch (e.g. Wolos 1991; Steffens and Winkel 1999). winner is the person showing the greatest skill. The majority of these prefer natural fisheries in Unfortunately this remains an anathema because urban and rural environments with an emphasis fish are rarely randomly distributed in water on enjoyment of nature, good fellowship and sim- bodies and drawing the best fishing station can ple pleasures. Considerable numbers of people, for have a major influence on the outcome of the example approximately 30% of persons fishing in competition. the UK, fish off beaches and rocky shores in pursuit of marine species such as codling (Gadus Game fishing morhua), whiting (Merlangius merlangus) or bass (Dicentrarchus labrax), which feed or breed in The origins of recreational fishing probably stem coastal waters. These fish are eaten. Leisure anglers from game fishing. It can be considered a ritualistic are not necessarily interested in catching large num- pursuit of fish species which put up a good fight. bers of fish or specimen fish, although this is consid- Target species in fresh waters typically include 370 Chapter 17 salmon and trout in temperate rivers, mahseer or an important role in supplementing diets. In tiger fish in tropical rivers and lakes, and marlin Eastern Europe, for example, carp is a traditional and sailfish in the sea. Proponents usually prefer to Christmas meal and many larger specimens are fly fish for salmon and trout, although artificial retained towards the end of the year for this pur- lures and natural baits are also used. In fresh pose. Similarly, in poorer countries the recrea- waters, these types of fisheries are usually highly tional catch supports domestic food supply. By regulated for access because they command high contrast, recreational fishing for a variety of economic and social values. Thus the participants freshwater species in Nordic countries, with com- tend to be from the higher income bracket of the mercial gears such as gill-nets, is exclusively for population. Similarly, in the seas, the cost of pur- consumption. Most fish caught from sea fisheries suing this type of fishing activity from dedicated are also retained for consumption. The contribu- boats makes it the domain of the wealthy. Never- tion to the landings from these fisheries is rarely theless, the contribution game fishing can make to accounted for in national statistics. However, the the local economy in terms of indirect expenditure fishing effort exerted by recreational fisheries can is immense (see Chapter 6.3). be massive, and it can account for a significant component of the catch for a country. For example, Mike and Cowx (1996) estimated that the recrea- Specimen and specialist anglers tional boat catch of the Trinidadian inshore A small proportion of anglers fish exclusively for fishery was equivalent to 11% of the total com- one species or specimen-sized individuals, mean- mercial catch, whilst the contribution of recrea- ing larger fish of that species. Specimen fishermen tional fishers to the total landings in the marine use rod and line, and target species that are long- inshore fishery of South Africa was over 30% lived and grow to a big size, for example carp (Cockcroft et al. 2000). In these cases the fish are (Cyprinus carpio) or pike (Esox lucius). Most spe- often sold in local markets and compete directly cialist fisheries are privately owned by individuals with the commercial fisheries. and syndicates who manage them for the par- ticular species in demand. They are particularly common in Western Europe, UK and France, and 17.2.3 Recreational fisheries attract anglers from long distances with the classified by location prospect of catching large fishes. Because fisher- Natural freshwater fisheries men want to improve on the size of fish caught, all fish are carefully handled and returned to the water Natural fisheries are those in which the productiv- after capture. These fisheries are more and more ity of the fishery or water body is not enhanced in being managed by stocking few, large fish in anti- any way. The proliferation of the stocks, and hence cipation that they will grow to a specimen size availability of fish for capture, relies entirely on quickly. natural reproduction and feeding. Such fisheries are bound by the strict limits of the productivity of the water in which they are practised, and by the Domestic consumption reproductive potential of the target stock. The pro- Although recreational fishing is considered a ductivity of the water is determined by geomor- leisure activity, a significant proportion of the phological considerations, including the nature of catch is removed for human consumption. This is the bedrock in the basin, the quantity and type of particularly true of freshwater game fishing, where effluent discharges from human activities, the the catch is considered to have high organoleptic depth and shoreline development of a lake, the qualities, especially salmon and trout in North extent of flooding in rivers and the residence time America and Scandinavia. However, in many in reservoirs. Reproductive potential is regulated countries of the world, recreational fisheries play by biotic factors such as abundance of the species, Recreational Fishing 371 the degree to which the population is harvested, the rate of drawdown in reservoirs or the extent of and competition and predation from other species flooding in rivers. (Fig. 17.1). It may also be influenced by physical These fisheries were traditionally the main tar- characteristics of the environment, such as the get of recreational anglers because they supported abundance of spawning substrates, temperature, good stocks of catchable fish and provided condi-

Factors increasing fish numbers

Illegal Accidental

Immigration

Planned Unplanned Natural recruitment

Stocking

FISH Pollution STOCKS Emigration

Accidental Stock Angling removal mortality Deliberate

Theft Cropping Disease Natural and mortality parasitism Cannibalism Predation

Old age Competition Other starvation Piscivorous Piscivorous animals fish birds

Factors decreasing fish numbers

Fig. 17.1 Factors influencing the numbers and size of fish in a fishery. (Source: adapted from Cowx 2001.) 372 Chapter 17 tions conducive to relaxation and enjoyment of is the stocking of artificially reared fish to boost nature, two of the main reasons why people go recruitment and production to overcome natural fishing (Wolos 1991; Roth et al. 2001). Unfortu- shortfalls. Stock enhancement can also be nately, in major industrial countries of Europe and achieved through the transfer of fish from one North America, many of the waters supporting water body, where the stock size is considered ade- these fisheries have suffered from a legacy of pollu- quate or at too high a density, to another. The main tion, habitat degradation, overfishing and flow ma- types of fisheries operated on the basis of stocking nipulation (Cowx and Welcomme 1998), and they are as follows. no longer satisfy the demands of anglers. Conse- 1 Put, grow and take. The commonest type of quently, anglers have turned more and more to stocked fishery is where juvenile fish are intro- highly managed systems (see below and North duced to grow on natural foods. The adult fish are 2002). Notwithstanding this, many good-quality later captured and removed from the system by natural fisheries still exist, especially game and recreational fishermen for consumption. The ex- coarse fisheries, but they are now considered tensive carp fisheries of Eastern Europe or the less productive in terms of quality of catch and large-mouth bass (Micropterus salmoides) fish- concerted action is needed to prevent further eries of North America fall into this category. loss. These fisheries are usually established because re- Natural fisheries are operated both on a catch- productive success of the wild fish population(s) is and-return basis, as in the coarse and centrachid impaired by there being too few mature fish or loss fisheries of Western Europe and North America of spawning and nursery areas. This type of stock- respectively, and on removal of part or all of the ing is only possible if the natural productivity of catch, for example game fisheries in North the water body provides good potential for growth. America and Europe and carp fisheries in Eastern Such fisheries are highly desirable to the managers Europe. In catch-and-return fisheries, emphasis is and owners because the high cost of growing fish orientated towards the welfare of the fish, in order to table size in a fish farm is borne by the natural to minimize fishing mortality and ensure that productivity of the water body. fish survive to be caught again, and to support 2 Put and take. Put-and-take fisheries aim to pro- the reproductive processes. In Western Europe the vide the angler with a quality fishing experience catch-and-release strategy often involves keeping by enhancing stocks on a regular basis. These fish- the fish in a keep-net for extended periods of eries are regularly stocked exclusively for the time until the fishing experience is complete (see capture and removal of fish by anglers. They 17.5.1). This is particularly true for match and classically target rainbow trout (Oncorhynchus leisure fishing for coarse fish. The catch-and- mykiss), brown trout (Salmo trutta), brook trout remove fisheries are managed to ensure sufficient (Salvelinus fontinalis) and lake trout (Salvelinus fish escape capture to replenish the stocks. spp.), although there is now considerable species diversification to meet the demands of anglers. Stocked fish are usually bred and grown on to a Stocked freshwater fisheries takeable size in a fish-farming facility and released When the quality of the fishing is below expecta- for capture. Successful put-and-take fisheries tion or does not satisfy the demands of the anglers, monitor the catch rates of anglers and restock on a fishery managers seek to intervene in the system continuous basis, but in the best-run fisheries to overcome the limiting factors. This problem stocking occurs when the fishing success falls arises principally because the fish stock is overex- below an established threshold (North 1983). The ploited, the productivity of the water is naturally time between release and capture is usually short, low, the habitat has been degraded or access to typically less than 4 weeks, although fishing pres- spawning areas has been compromised and re- sure often increases around the time of stocking cruitment is poor. The principal intervention used and many fish are removed within a few days of Recreational Fishing 373 release. Put-and-take fisheries are practised in as gill-nets to catch fish, although this, as in fresh some natural waters, such as trout streams, but waters (Section 17.2.1), is more akin to subsistence more normally are restricted to purpose-built fishing than recreational angling. With the ponds or reservoirs, often associated with a fish exception of game fishing, the catch is usually culture station (Templeton 1995; Flickinger et al. for domestic consumption, although some com- 1999). These types of fisheries are generally petitions are organized. Sea fishing is often highly operated as commercial ventures, so are dependent seasonal, exploiting the migratory patterns of on anglers fishing the water, and paying a fee for species as they move inshore to breed or feed. This the fishing experience and removal of a number of is mainly an open-access and unregulated activity, fish for consumption. To prevent overfishing or to with the exception of a few gear restrictions. How- regulate successful anglers removing too many ever, in recent years access restrictions and closed fish, most fisheries have a bag limit. In successful seasons have been imposed on certain fisheries fisheries the density of fish stocked is usually which exhibit signs of overfishing (Gautman and high and scope for any natural production is Hicks 1999). limited. 3 Put, catch and return. There is an increasing trend in recreational catch-and-return fisheries 17.3 RECREATIONAL in Europe and North America to enhance the fish FISHERY ASSESSMENT stocks to increase angler satisfaction and demand METHODS (Kohler and Hubert 1999; North 2002). These types of fisheries either: manipulate the fish communi- An integral part of the management of recreational ties to provide specialist fisheries for a particular fisheries is an understanding of the status of target species; stock specimen fish; or intensively the resources in terms of fish stock structure, the stock water bodies of target species well above value of fishing to the anglers, and to the local and that sustainable by natural productivity. The fish- regional economies, and, perhaps most impor- eries are intensively managed to maintain the tantly, of end-user satisfaction. This information desired stocks and ensure angling success is high. is required both to formulate management However, there is now considerable debate over measures and to assess the impact of management the general welfare of the fish in intensively policy on the fishery performance. For exploited stocked fisheries because the fish are living in natural stocks, annual assessment of the stock highly stressed conditions and exposed to the status through catch-and-effort sampling appears further stresses from capture, handling and release acceptable, whereas put-and-take fisheries need (Section 17.5.1). Specialist and specimen fisheries weekly monitoring for angler satisfaction and are, by contrast, considered acceptable because stock depletion, so that management can respond they are usually stocked at low density and cater quickly to any measured deterioration in either for anglers who seek to catch large individual fish characteristic. For unexploited stocks in catch- or specific species. They are also highly managed and-return fisheries, trends are usually the main and often have restricted access. requirement. Surveys to collect this type of information are generally constrained by money and manpower, Sea fisheries and the degree of precision required in the infor- Recreational sea fishing almost exclusively targets mation collected to meet management objectives wild stocks in usually one of three ways: rocky (Cowx 1996). Problems are also encountered in shore fishing or beach casting; inshore line fishing; the assessment of the economic status because of and offshore sport/game fishing for oceanic the difficulty of putting value on non-tangible species. In some Scandinavian countries, recrea- resources. Several methods have now been devel- tional fishermen also use commercial gears such oped to overcome these problems and provide 374 Chapter 17 baseline information on which management Moore 1998). However, it should be recognized policy can be formulated. that each approach has distinct advantages and Two distinct strategies for obtaining the infor- biases over the other (see Table 17.1). mation from anglers are used (Malvestuto 1983; Off-site event recall methods tend to make use Pollock et al. 1994; Cowx 1996): of lists of names of anglers, perhaps those in receipt • off-site event recall methods such as log books, of a fishing licence or permit, or boat licence, from mail questionnaires and phone surveys, which which a random sample is selected to provide the require the angler to evaluate the fishing experi- necessary information, or they make use of obliga- ence some time after the event (e.g. Axford 1991; tory returns of information linked to the issuing Beaumont et al. 1991; Bunt 1991; Churchward and of licences. Mail and telephone surveys tend to Hickley 1991; Cowx 1991; Gardiner 1991; Gerard be low cost and can have a regional coverage, but and Timmermans 1991; O’Hara and Williams suffer from biases associated with non-response, 1991; Small 1991). vagaries of recall and memory, angling prestige and • on-site intercept methods such as aerial count enthusiasm (Bunt 1991; Churchward and Hickley methods, access point methods and roving creel 1991; Dekker 1991; Gardiner 1991; Löffler 1991; methods, which collect information at the time of Small 1991). the event or immediately following it (e.g. Cryer Dependence on lists of names also has biases: and MacLean 1991; Gerard and Timmermans for example, not all anglers may be required to 1991; Malvestuto 1991; Pawson 1991). have licences, especially when sea fishing. To Information collected by either method can in- obtain acceptable data, the recall time interval clude total weight of catch, weight or number of should be kept to a minimum (<2 months pre- each fish species caught, time of day spent fishing, ferred) and the questions need to be simple. If at all duration of fishing experience in terms of number possible, the participants should be encouraged to of man-hours or other unit of effort, location of complete the questionnaires at the time of the fishing and method of fishing. These data are used fishing experience, i.e. set up a fishing diary or to determine the catch per unit of effort, usually logbook, although these too can be biased. Licence represented as g man-hr-1, fish man-hr-1 or similar returns overcome some of the problems of random measure. In addition, the survey methods can be sampling but, despite statutory obligations, still used to collect information to determine the value suffer from non-response and recall biases. The of the fishery, using contingent valuation method- non-response can be minimized by sending out ologies or other similar techniques (Pollock et al. reminders, but rarely is a 100% return rate 1994; Gautam and Steinback 1998; Postle and achieved.

Table 17.1 Advantages and disadvantages associated with event recall and on-site intercept approaches for recreational fishery assessment.

Approach Advantages Disadvantages

Event recall methods Low cost Non-response Regional coverage Recall bias Immediate response (phone surveys) Digit bias Prestige bias Avidity bias On-site intercept methods Minimization of response biases High cost Visual assessment of information exchange Biases based on survey design Interruption of angling experience Recreational Fishing 375

On-site intercept methods minimize the biases tion of management programmes. Lorenz curves, associated with event recall methods because they which quantify the probability of catching a certain have a high response rate, they allow direct infor- number or weight of fish (O’Hara and Williams mation exchange which can be confirmed by the 1991), or the utility-per-recruit concept of Die et al. interviewer, and anglers are not required to recall (1988), which uses an objective measure of angler catch information. Disadvantages are the high satisfaction in place of catch weight in the usual cost per angler interview, difficulties in contacting yield-per-recruit model (Shepherd and Pope, Chapter a representative sample of anglers in a large geo- 8, this volume), are good examples of this approach. graphical area, and reliance on a random stratified • The use of catch per unit effort (CPUE) as an survey design, similar to that used for commercial index of change in stock abundance for recrea- fisheries, which may introduce biases associated tional resources must be evaluated, particularly with the estimator used. With roving surveys, the for those fisheries where fluctuations in stock size interviewer contacts the anglers as he moves and high rates of exploitation are critical factors around the fishing area along a predetermined related to efficient functioning and the economic route. The primary weakness of roving survey value of the fishery (put-and-take systems). methods is that catch-and-effort information is Many of the methods introduced above tradition- based on incomplete rather than complete fishing ally focus on estimation of the biological para- trips, because anglers are contacted while they are meters, i.e. catch, effort and CPUE related to fishing. In access point surveys, the interviewer is fishery. However, in recent years there has been an stationed at a strategic site such as a car park, jetty increasing need for valuation of the economic and or boat landing beach used by anglers, preferably at social benefits of recreational fisheries to justify the end of the fishing trip. The surveyor then ran- their position in a wider multiple resource-user domly selects a predetermined number of anglers environment (Cowx 1998b). As already indicated, for questioning. The access point survey is ideal this is a complex problem because it is difficult to where the anglers must leave via a small number of put value on non-tangible recreational resources. points or where anglers report their catches at a Nevertheless several approaches, which can be central location. Although the complexity of the built into the traditional angler intercept and questions in on-site methods can be increased it event recall methods, have been developed to should not be so great as to interfere with the overcome these problems (Table 17.2). angler experience. On-site methods can also be Reviews on how to use these approaches are combined with aerial surveys, boat counts or bank given by Pollock et al. (1994) and Weithman (1999). angler counts to determine the total fishing effort However, in general terms, the gross economic and assess the actual exploitation patterns. value of a recreational fishery is composed of: (1) To ensure accuracy of the data retrieval, several explicit costs of local, regional and national ser- important aspects should be addressed (Cowx vices which contribute directly to the angler; (2) 2001). total angler expenditures directed towards the • Event recall methods must be validated with fishery, derived from licence and permit fees, respect to the accuracy of the responses. Calibra- equipment and travel costs; (3) the value over and tion with on-site survey data is one option. above the actual expenditures that anglers would • On-site intercept methods must be evaluated be willing to spend to fish; and (4) the additional with respect to field techniques and design effi- value based on non-users’ willingness to pay to ciency. New design alternatives must be devel- preserve the fishery (Malvestuto 1983; Weithman oped, particularly with respect to obtaining 1999). On-site intercept and event recall methods precise estimates of catch rate. can be used to provide estimates of average expen- • Indices which truly reflect the quality of the diture per angler-trip or over a defined period, by fishery from the angler’s point of view must be asking the appropriate questions. The same sur- developed, and used for justification and evalua- veys can be used to provide information on the net 376 Chapter 17

Table 17.2 Methods of assessing fishery benefits. (Source: after Weithman 1999.)

Benefit measurement Approach

Non-monetary Social well-being Angler survey on changes in a fishery Psychophysical measures Angler survey on aesthetic appeal Multi-attribute choice approach Angler survey on a variety of fishery characteristics Attitude measures Angler survey of factors that affect the quality of fishing Social impact assessment Projection of changes that will likely result from a new policy Monetary Economic impact assessment Angler expenditure used as input for regional economic models Economic value assessment Angler expenditure used as input for travel cost or contingent valuation Total economic valuation Angler expenditure used as input for contingent valuation, plus non-consumptive user values Combination Total value assessment Comprehensive evaluation that combines non-monetary and monetary evaluations to determine social and economic values

willingness to pay, often defined as consumer Table 17.3 Interventions used in the management of surplus, using contingent valuation or travel cost recreational fisheries. methodologies (Pollock et al. 1994; Baker and Fish stock Pierce 1997; Postle and Moore 1998). This will pro- Restocking to enhance stock vide an economic value of angling to the region or Introductions to increase species diversity the value individuals place on the fishery, and they Culling to remove predators and pest species can be used as persuasive arguments for the main- Fishery tenance and improvement of the fishery. Closed seasons Closed areas Access rights 17.4 MANAGEMENT OF Gear restrictions RECREATIONAL FISHERIES Harvesting size regulations Habitat Recreational fisheries are underpinned by a com- Rehabilitation of habitat plex interaction of physical, chemical and biologi- Increasing fish habitat diversity Improvement of water quality cal conditions, which need to be regulated in such a manner as to enhance the fishery output and maintain quality fishing (Fig. 17.1). Management of recreational fisheries is realized in three ways, Stock enhancement targeting the fish, the fishery and the habitat (Table 17.3). Stocking is probably the most widespread, and abused, management tool used in recreational 17.4.1 Stock manipulation inland fisheries today (Cowx 1998a). There is little evidence of successful enhancement of marine fish Management of the fish population size or com- stocks through stocking (Howell et al. 1999), and munity structure is based on a variety of stock thus it is a technique rarely considered for enhanc- manipulation techniques (Fig. 17.1). ing recreational marine fisheries. Most countries Recreational Fishing 377 report stocking of freshwater fisheries as more of takeable-sized fish for easy capture in put-and- conventional approaches to management have take fisheries, or a large number of juveniles which failed to control fisheries exploitation or reduc- are intended to grow on to a large size and satisfy tion in stock biomass through environmental the demands of the angler. degradation. The importance of stocking to sup- There are a variety of strategies used to ensure port recreational fisheries is demonstrated by the the success of a stocking exercise in inland waters huge stocking programmes that take place each (Cowx 1994a). Explanation of these strategies is year (see Table 17.4), and the economic value of the beyond the scope of this chapter, but issues that activity was illustrated by Cowx and Godkin must be addressed are: sources of stocked fish; (2000), who estimated some US$90 million of fish species to be stocked; stocking density; age and were produced in aquaculture units in the Euro- size of fish to be stocked; choice of season for stock- pean Union countries in 1997. This represents an ing; acclimation of stocked fish; mechanism of estimated 7.6% of the value of inland fisheries stocking; potential interaction with wild fishes; and aquaculture production in these countries, transmission of parasites and disease; and genetic and does not account for the value-added revenue integrity of the wild stocks (see Cowx 1994a for generated by the fishing activity itself. details). Unfortunately, this strategy to enhance the fish stocks is often used to respond to anglers’ com- Increasing species diversity plaints that the fishing is poor or just to increase the fish stocks in general. Stocking in natural One of the common complaints about recreational waters should aim to improve recruitment, bias fisheries is poor diversity of species. To overcome fish assemblage structure to favoured species, or this problem there has been considerable move- maintain productive species that would not breed ment of fish from one water to another (trans- naturally in the system (Cowx 1994a). However, locations) or the introduction of exotic species all this should be carried out so that there is no throughout the world (Welcomme 1988, 1992, impact on the indigenous fish populations (see 1996; Cowx 1997, 1998a; also see Juanes et al., Section 17.5.2). In man-made water bodies stock- Chapter 12, Volume 1 and Section 17.5.2 for more ing is usually carried out to create fisheries discussion). The most commonly introduced or which satisfy the general demand of the public, for translocated species involve cyprinids, with 57 example to create specialist fisheries for trout, species put into European fresh waters, of which carp or large-mouthed bass. Irrespective of the type common carp is the most frequently involved; of water body, the technique involves the injection salmonids with rainbow trout, brook trout and

Table 17.4 Numbers of fish (millions) of different species stocked in fresh waters in North America and the European Union countries in 1998.

Eggs Fry Fingerlings Sub-adults Total

United States and Canada Salmonidae 34999 24691 149363 67188 276241 Warm-water sport fishes 24581 112523 4030 141134 Cool-water sport fishes (Percidae and Esocidae) 4502 1193705 71028 169 1269404 European Community Atlantic salmon 3124 13732 2559 1077 20492 Salmonidae (all) 214372 Non-salmonidae (coarse fishes) 210675 210675 378 Chapter 17 brown trout being the most common and, finally, This technique has rarely proved successful as it largemouth bass (Micropterus salmoides). All can create an imbalance in the fishery, which these species groups are to enhance sport fishing. becomes dominated by large numbers of small, Although the movement of fish is highly regu- often planktivorous, fish (Templeton 1995; Pers- lated in most countries, indiscriminate actions are son, Chapter 15, Volume 1; Kaiser and Jennings, commonplace and lead to introduction of diseases, Chapter 16, this volume). The removal of the loss of genetic integrity and species extinctions in predators can also reduce predation pressure on the the worst possible case (Cowx 1997, 1998a; Cowx faster growing and prolific breeding species, which and Godkin 2000). Introductions have also arisen in many cases may not be the target of the anglers, because of release or escape of fish used as live bait and can ultimately be a nuisance. Furthermore, for predatory species such as pike (Welcomme fish predators often control their own numbers 1988; Cowx and Godkin 2000). Such releases form through cannibalism, and the removal of larger more than 1% of cases. individuals from a population can leave many smaller conspecifics which have a great impact in terms of reduction in prey density. In marine Elimination of unwanted species ecosystems, there is evidence of mixed effects Elimination of unwanted species that either com- of reductions in predatory fishes (Kaiser and pete with, or prey upon, target species is another Jennings, Chapter 16, this volume). tool often used in recreational inland fisheries. Of more concern at present is the control of fish- Cropping can either target excessive numbers of eating birds, particularly cormorants (Phalacroco- fish of a species or species group to allow the re- rax sp.) and herons (Ardea sp.). In recent years, mainder to grow to a larger size, or can target pre- their numbers have increased dramatically at a dators, which are considered undesirable because rate of around 16% p.a. to about 150000 pairs be- they prey on the fish targeted by anglers. Cropping tween 1978 and 1993 around inland waters in is a useful tool in fisheries that have a high stock Europe and North America (Russell et al. 1996). density or biomass and the growth of the fish is They are perceived to cause considerable damage poor. The culling of predators has been used with to fish stocks. Whilst there is no doubt they can some success on fisheries where new species of cause damage to intensively stocked still-water predators have been introduced deliberately or fisheries, there is still debate over the damage to accidentally and have caused the demise of the other natural freshwater fisheries (Feltham et al. indigenous stocks. An example is the control of 1999). For example, at least 57% of carp stocked pike perch (Stizostedion lucioperca) in the into a still-water fishery near London were con- Midlands of England (Smith et al. 1998). However, sumed over a period of 4 weeks (Feltham et al. there is considerable debate over whether the 1999) and up to 75% of stocks can be removed method is inappropriate for controlling indigenous from fish farms by cormorants (Russell et al. 1996). predator populations (see next section). Consequently, there are calls for control of birds, especially cormorants, by fisheries managers and anglers, but this is hotly contested by conserva- Predator control tionists (Russell et al. 1996). Notwithstanding, Predator control is perhaps one of the most control measures for cormorants, such as using controversial methods of fishery management. scaring noise or visual methods or shooting small It involves regulating the numbers of predators, numbers of birds, have proved rather ineffective, usually fish or avian, to reduce predation pressure. and the creation of refuge habitats and stocking The culling of piscivorous fish is a typical example with larger fish at times when the birds are of this approach, with many fisheries having a foraging off the coast are recommended to mini- policy to kill major predators if they are caught. mize the impact (McKay et al. 1999). Recreational Fishing 379

17.4.2 Fishery regulations cause the season does not always coincide with the reproductive period, as is typically observed in the In addition to direct intervention on the fish UK (Hickley et al. 1995). Furthermore, there is no populations/communities, recreational fisheries evidence that closed seasons lead to better recruit- also require the introduction and enforcement of ment within the fish populations, as other biotic various regulatory constraints to prevent over- and abiotic factors appear to have a stronger influ- exploitation of the fishery and maintenance of a ence on this process (Mills and Mann 1985). suitable stock structure. The various measures Closed areas are designed to protect stocks that are commonly operated in recreational fish- directly by denying access to the angler. These eries through Europe (Hickley et al. 1995) and can range from sanctuary areas, where fishing is North America (Noble and Jones 1999), and their prohibited to protect vulnerable life stages or expected outcome, are summarized in Table 17.5. species of fish (e.g. Polunin, Chapter 14, this These regulations are similar to those imposed on volume), to restrictions on fishing in areas where major commercial fisheries but the scale and in- the fish are particularly vulnerable to exploitation, tensity are usually less draconian. Elsewhere in such as the aggregation of salmon below weirs the world, regulations are mainly restricted to the and waterfalls during their upstream migration. gear that can be used. For example, Ontario, Canada, has used seasonal Closed seasons are imposed to protect the fish sanctuaries to restrict anglers from large-mouth mainly during the breeding season or the early bass spawning areas for over 60 years (Noble and development stages. This includes protection of Jones 1999). Management of put-and-take fisheries migratory fish as they move towards spawning or may include closed areas where stocking takes feeding grounds, especially at places where they place to prevent anglers removing fish when they will be particularly vulnerable to heavy exploita- are highly vulnerable immediately after they have tion, and provides a respite for fish to spawn been stocked. unimpeded (Noble and Jones 1999). In practice Where the catch is removed for consumption, this action has been extended to protect stocks limits are frequently placed on total catch to pre- which are heavily exploited and thus restrict vent overexploitation and conserve the spawning catch, for example chinook salmon (On- stock. Such restrictions allow for the sharing of the corhynchus tshawytscha) off the west coast of catch when stocks are low or under intense fishing Canada (Walters and Cox 1999). This restriction pressure. Bag limits are commonly applied in mi- has often come under heavy criticism because gratory game and put-and-take fisheries to ensure closed seasons are wrongly timed and do not pro- equity of catch. An alternative to the bag limit is tect the fish when they are most vulnerable, be- catch and release (see Section 18.2.3) where restric-

Table 17.5 Techniques for regulating angling effort and the ecological requirements of the fish stocks that are addressed.

Regulatory Population Broodstock Undisturbed Free passage Fish welfare technique size protection spawning

Closed areas * * * Closed season * * * * Catch limit * Fishing pressure * * Type of gear * Size of fish * * * 380 Chapter 17 tions are placed on catch and all fish must be re- mortality rates. In addition, problems may arise leased back to the water. However, for this action because too little is known about post-release to be successful it is important that proper proce- mortality (e.g. Bettoli and Osborne 1998; Cooke dures are followed with regard to playing, handling et al. 2000; also see Section 17.5.1). and releasing of fish. A combination of both catch restrictions and release is often employed so that the angler can still enjoy the experience if fishing is 17.4.3 Habitat management good but he has reached the bag limit. Fisheries managers use habitat management tech- Gear restrictions are used to reduce exploita- niques to improve access to, and quality of, fishing, tion of populations by influencing the efficiency of improve degraded habitats and increase the pro- the fishing method, the species caught or the size duction potential of the fishery. The following of fish caught (Misund et al., Chapter 2, this vol- are some of the main actions used: construction of ume). There is also the requirement to protect indi- groynes and fishing platforms; creation of fishing vidual fish by reducing any stress and physical pools; raising of water level using small dams, damage associated with being caught and held in cutting off access to spawning streams to control keep-nets, if the intention is to return them alive. recruitment; blocking of outflows (usually with In some cases, angling may still impose physiolog- some kind of screen) to prevent stocked fish escap- ical costs on the fish, even if the fish are released ing; removal of natural barriers to fish migration (Forsgren et al., Chapter 10, Volume 1). It is neces- such as waterfalls; provision of fish ladders; sary to base the regulations on sound information addition of fertilizers; liming; clearance of aquatic on the life histories and effectiveness of different weeds by herbicides or the use of grass carp; and angling techniques, but this information is rarely cutting of bankside vegetation including trees (see available. In recreational fisheries, angling is Cowx and Welcomme 1998 for details). All these often restricted to hook-and-line fishing practices. are designed to improve the fishing experience and Commercial gears, such as nets and traps, are if carried out in a sympathetic manner will rarely used except in countries where recreational enhance the status of the fishery. fishing has more in common with subsistence Engineering of the environment to improve fishing and provides food for consumption. levels of reproduction, shelter, food resources and The final regulatory mechanism commonly vital habitat is a tool which deserves considerably used is limitation on the size of fish that can be more attention. It is a long-term solution to taken. Restrictions of the size of fish that can be improving fish stocks, but is expensive, and unless removed for consumption are commonplace in carried out correctly can cause further degradation commercial fisheries and are equally applicable to of the habitat (Cowx and Welcomme 1998). It is recreational fisheries. The restriction is designed recommended that full consultation with water to ensure all immature fish are returned to the resource managers and environmental experts is water to allow a self-sustaining population to made before any action is taken, to avoid undesir- persist. This is sometimes supported by restricting able responses. retention of larger mature individuals to allow spawning escapement, thus creating what is termed a ‘slot size’ for fish that may be retained. These restrictions generally refer to salmonid 17.5 ISSUES RELATING TO fisheries in Europe (Templeton 1995) and North THE DEVELOPMENT OF America, and work especially well where harvest RECREATIONAL FISHERIES is high and recruitment low (Noble and Jones 1999). For size limitations to work, they must be Despite the importance of recreational fisheries based on sound information about population worldwide, there is a general perception that size structure, size at sexual maturity and natural natural fisheries have undergone major, often Recreational Fishing 381 adverse, changes in recent years. If recreational planned, taking into account wider environ- fisheries are to be sustained and developed for fu- mental considerations (Cowx and Welcomme ture generations, there are a number of issues that 1998). need resolution. These can be divided into three Litter. Angling, like other human activities, main categories relating to: (1) angling practises creates litter, including everyday items such as per se; (2) stock enhancement activities, which are drink cans, polythene bags and paper cartons. potentially damaging to the aquatic ecosystem Angling litter is, however, characterized by the in- (Maitland 1995); and (3) cross-sectoral interactions. clusion of items such as discarded bait containers, lead shot, nylon line and fish hooks (Cryer et al. 17.5.1 Angling practices 1987a). The latter, which are either discarded un- intentionally or lost inadvertently during angling, Disturbance and damage. Angling, although are particularly damaging, because birds and other essentially a quiet and often solitary activity, animal life become entangled. Lead shot has been can result in disturbance to wildlife. Commonly, implicated in the decline of bird species such as waterfowl, and coastal and wetland birds, many of swans (Cygnus olor), because they inadvertently which are now rare, are liable to disturbance if ingest discarded items when feeding. This problem access of anglers to waters or shoreline is uncon- has largely been addressed through the introduc- trolled (Cryer et al. 1987b). Most damage is done tion of non-lead alternatives in angling. at the nesting time, when birds are disrupted on Groundbaiting. The practice of groundbaiting their nests or prevented from gaining access to with cereals, maggots or other bait is commonly their nests (Maitland and Lyle 1992). There are also practised in Europe and North America to attract many mammals commonly found associated with fish to the hook. When used excessively, it can lead the rivers and lakes, most of which are shy, such as to a deterioration in water quality (Cryer and the otter (Lutra lutra), and sensitive to distur- Edwards 1987; Edwards 1990), increase phospho- bance. In addition they prefer secure places to rear rus loading (Edwards and Fouracre 1983), and lead their young (Jefferies 1987). Closed seasons or to a substantial reduction in benthic fauna (Cryer protected areas are designed to minimize these and Edwards 1987). In many places this practice is impacts, but problems still persist. now discouraged to minimize the problems. Disturbance is also caused by noise and pollu- There is a surprising lack of information on the tion, such as oil leaks from boat engines where ways in which stocked systems function. The few fishing takes place from a boat. Boat wakes can also data sets that exist (see Cowx 1998a for examples) cause erosion of river banks, especially where show interplay of two main variables: area of the movements are excessive and uncontrolled stocked system and stocking rates. They indicate (Pygott et al. 1990; Ellis 1998). This leads to col- that the output of stocking is highly variable and lapse of banks, loss of riparian vegetation, and, on a that in the majority of cases stocking does not moresubtlelevel, change of littoral water tempera- achieve any major increase in fish production or tures, which directly affects juvenile fish growth yield (Cowx 1998a). Indeed, anecdotal evidence and recruitment (Hodgson and Eaton 2000). suggests that the majority of stocking in rivers to Habitat management. Anglers also cause enhance fish populations is of marginal value. The physical damage to the habitat, especially riparian notable exceptions are the stocking of migratory vegetation, to gain access to the water. This can salmonids, especially where the river has been lead to loss of sensitive wetland flora and fauna if heavily degraded, such as in Sweden, or the left uncontrolled and can damage the integrity of polluted rivers of Western Europe, such as the the fishery (Swales and O’Hara 1983; Cowx 1994b). Rhine and the Thames, once water quality had Other management practices, as described in been improved. Improvements in catches of non- Section 17.4.2, can also be damaging; therefore to migratory salmonids have been achieved by stock- minimize any problems they should be carefully ing rainbow trout or brown trout, but these results 382 Chapter 17 tend to be short term because the recruitment presented by environmental lobby groups is that bottleneck has not been removed or, as in the case fish may sustain damage, especially from barbed of rainbow trout, the species does not breed suc- hooks, which increases their proneness to disease cessfully in European conditions. and feeding difficulties. As a consequence, some The stocking of rivers with cyprinids has regions of Europe have now banned put-and-take achieved little success, considering the large num- fisheries, and the use of live bait and keep-nets bers that have been introduced. For example, (e.g. Norway, Netherlands and several Länder in after reviewing over 50 years of stocking activity Germany), and others are looking carefully at in the lower Welsh Dee, Pearce (1983) concluded the issue (Wortley 1995). Whatever the outcome, that ‘stocking on the Dee would not appear to anglers must be aware of animal welfare issues and have been successful, only extremely limited continue to do everything possible to minimize short-term and scarcely any long-term benefits to the impact of the activity on fisheries and wildlife. angling or stock recruitment have resulted’. This is despite there having been some 76 stocking 17.5.2 Stock enhancement events over the study period, involving approxi- mately 400000 fish with, typically, two or three As previously indicated, stocking, which is used species being used on each occasion. extensively to enhance fisheries, is frequently car- The stocking of still waters exhibits the ried out with no due regard for the environmental opposite outcome, especially when stocked with or ecological consequences. Stocking can be dam- robust species like rainbow trout and carp. These aging to the native stocks through competition, have been highly successful and constitute fishery predation, loss of genetic integrity, or the spread of enhancement in a greater proportion of cases. disease and parasites (Cowx 1994a, 1998a; Cowx For example, the numerous put-and-take and and Godkin 2000). Unfortunately there is a rela- catch-and-return fisheries found throughout tively little information about the effects, includ- Europe are based on this premise, and they have ing successes, of various stocking practices (Cowx proved highly successful. Intensive carp fisheries 1998a). If the species is being released in high num- and numerous other cyprinid and specialist fish- bers then changes in the ecosystem are likely to eries are supported in the same way. However, occur via fish species interactions and food web stocking is not always successful, and the few dynamics (see Persson, Chapter 15, Volume 1, and reported examples of where attempts have been Pauly and Christensen, Chapter 10, this volume, made to support commercial fisheries based on for commentary on ecosystem responses to pertur- stocking suggest these are not worthwhile bation). Vulnerable fish species, and aquatic flora (Salojärvi and Mutenia 1994; Löffler 1998). and fauna, can be eliminated through predation, Animal welfare. There is growing concern that and the stocking, intended to enhance the fishery, the holding of fish at high density in keep-nets, can result in the opposite effect, including elimi- coupled with the hooking, playing and handling nation of species (Cowx 1997). The loss of genetic of the captured fish, causes unnecessary distress integrity of the native stock, which is thought to (Berg and Rösch 1998). Although the impact of have adapted to local environmental conditions catch and return on fish behaviour and populations over many years, is a major issue and is considered is not well understood, there is evidence that to be responsible for a decline of many fisheries, fish do suffer from being caught and handled, especially salmon stocks (Carvalho and Cross procedures which have reduced recruitment suc- 1998). cess (e.g. Bettoli and Osborne 1998; Cooke et al. The introduction of new species to promote 2000). By contrast, recent studies on holding fish in angling diversity was common practice worldwide keep-nets suggest that the fish are not unduly in the 1960s and 1970s (Welcomme 1988), but con- stressed until the density held is high (Pottinger cerns over the impact of this activity and the im- 1997; Raat et al. 1997). Another argument being plementation of tough regulations has restricted Recreational Fishing 383 the practice in industrialized countries in recent first instance. In addition, there are many examples years. Introductions, for example rainbow trout where fish species have been illegally or acciden- and largemouth bass, have proved successful in a tally introduced to increase the diversity of the number of cases but this has usually been at great target species of anglers. Finally, there are many cost, most often expressed through the demise of cases of fish introductions stemming from the use indigenous species or the spread of diseases (Cowx of live bait for catching predators (Cowx 1998a). All 1997, 1998a). Other introductions have been disas- these practices are potentially damaging to fish- trous, with wholesale ecosystem change and elim- eries and have rarely proved successful (Cowx 1997). ination of species (Cowx 1997). With the recent Indeed, they usually lead to the demise of the resi- decline in the status of many fisheries, there is dent fish populations and further deterioration of renewed interest in species introductions. This the stocks. Anglers and managers alike need to be should be prevented at all costs, and the causes of educated in the problems and consequences associ- the deterioration in extant fisheries should be ated with stocking and introduction of fish, whether identified and where possible addressed. If a deliberate or accidental, and all efforts should be species is to be considered for introduction, it made to minimize this management practice. should be carried out under the auspices of the appropriate governmental organization and 17.5.3 Subsector interactions following guidelines such as those promoted by the United Nations Food and Agriculture Organi- One of the major constraints on the future develop- zation (FAO 1996), the International Council for ment of recreational fisheries is from other users the Exploration of the Seas (ICES 1988) or the and occupiers of the water body. Such activity can European Inland Fisheries Advisory Committee be classified as cross-sectoral interactions and (EIFAC, see Coates 1998). mostly arise from conflicts of interest. Within the In reality, stock enhancement practices are development of any activity there is a need to con- often carried out in response to angling pressure sider the overall framework within which the and do not address the underlying problems within activity exists (Fig. 17.2). There is no value in the fishery, such as recruitment bottlenecks or attempting to improve recreational fishing in a overfishing. In this respect stocking exercises are multi-user situation if cross-sectoral issues are not just short-term fixes and maintenance of stock addressed. These issues arise from interactions: (1) size will require continuous intervention. It is between commercial and recreational fishing; (2) thus not necessarily a desirable solution and alter- within recreation fishing groups; and (3) with other native mechanisms should always be sought in the aquatic resource users. Furthermore, recreational

Water resource Industrial development Recreation development- pollution

Conservation Urban development RECREATIONAL FISHERIES Fish-eating birds

Navigation Water quality

Commercial Agricultural Flood defence Fig. 17.2 Factors affecting fisheries recreational fisheries development. development 384 Chapter 17

fishing cannot be treated as an open-access activity damming for power generation and water supply, because it is usually governed by considerable leg- flood alleviation works, weirs, intakes, bridges and islative and administrative protocols. This is often similar structures (Fig. 17.2; Petts 1984; Cowx and tied up with rights to fish, and regulations govern- Welcomme 1998; Kohler and Hubert 1999). All of ing such issues at closed seasons, gear and catch these activities have resulted in a shift in the status restrictions, and licensing, most of which are set of the fisheries and a general decline in the yield. In up to protect the fishing, but others to protect the these circumstances fisheries are not considered of environment. Access is also further complicated sufficiently high priority or value, and thus suffer by social issues such as traditional-use rights and in the face of economically and socially higher family obligations. priorities, for example agriculture, hydroelectric Commercial and recreation fishing interac- power production or flood prevention. tions. Direct conflicts exist between commercial Problems relating to pollution are in the main and recreational angling because they exploit the being addressed in Western industrial countries, same resource base. The arguments usually relate and waters that were once fishless or supported to commercial fisheries depleting stocks through poor stocks are now recovering. Such water quality overfishing, and demands from the recreation improvements are, however, rare in Third World sector to restrict commercial exploitation, countries, where the scenario remains one of although many studies indicate that commercial continued environmental degradation and conse- and recreational fisheries can coexist (see Hickley quent demise of the fisheries. Notwithstanding, and Tompkins 1998 for examples), supplementing water quality improvements do not always lead to each other, creating an overall larger output than a desirable outcome in terms of fishery status. The would have resulted from sport fishing only. When cleaning up of water bodies often results in reduced the commercial and recreational fisheries inter- nutrient input and the loss of productivity, which fere with each other, the allocation of the harvest has a knock-on effect on the fish community generally falls in favour of recreational fishing, as it structure and stock size. This results in a change is perceived to have an overall greater benefit to in the catch composition and reduction in catches, society (Cowx 2001). which is quickly followed by complaints from Interactions between recreation fishing groups. anglers (Cowx 1991). There is also increasing conflict between groups of Abstraction of surface or ground water, or the recreational fishermen, for instance between resi- impoundment of rivers, may also damage fisheries dents and non-residents (e.g. Morgan 1999). Rea- through reduced flows downstream of reservoirs, sons for this include: the loss of access to fishing the interruption of migratory pathways and the waters resulting in crowding problems and over- isolation and drowning of spawning and rearing exploitation on the remaining waters; increased areas. Conversely, reservoirs may provide attrac- specialization among anglers, which puts pressure tive fishing opportunities, especially when inten- on attractive waters; and increased focus on tourist sively managed. Resolution of the conflicts caused fishing in many areas, so introducing new angler by the structures is difficult and costly because groups to many districts, which may cause differ- of the high social and economic value of water ent conflicts of interest related to social and alloca- resource schemes. tion controversies (Hickley and Tompkins 1998). Interactions between other recreational uses of Cross-sectoral interactions. Perhaps the great- water, as well as recreation in the vicinity of water est threats to recreational fisheries come from bodies, are also sources of conflict. The following outside the fisheries sector. Aquatic resources are pursuits can impact on recreational fisheries: bird subject to numerous anthropogenic perturbations, watching, cycling, motor boating, canoeing, cruis- such as pollution discharge from agricultural, ing, rowing, sailing, diving, swimming and domestic and industrial sources, eutrophication, bathing, water-skiing, wind surfing, water bikes acidification, afforestation, mineral extraction, and wild fowling. The problems arise from distur- Recreational Fishing 385 bance of fish, noise, litter, the loss of access to the Furthermore, increasing pressures on aquatic water body and general disruption of the angling resources dictate that fisheries can no longer be experience (for examples see Pygott et al. 1990; treated in isolation, and an integrated approach to Ellis 1998; and Gerard 1999). aquatic resource management is required (Cowx 1998b). Consequently, the way forward for the management of recreational fisheries must be as a 17.6 VALUE stakeholder in the wider resource-user domain. It is essential that the fisheries are well represented From the previous section, it should be recognized in any decision making and that their true value is that recreational fishing activities and develop- promoted to the full. ment in a multiple-user environment are fraught with problems. Fisheries are often considered mar- ginal activities because the value of the resource is 17.7 CONCLUSIONS usually ill defined and poorly represented from an economic and social perspective. Fisheries are Increasing pressures on aquatic resources dictate traditionally managed based on the quality of the that fisheries can no longer be treated in isolation fishing experience, whereby the managers use that and an integrated approach to aquatic re- simple techniques such as stocking to maintain source management is required (Cowx 1998b). As fishery quality. However, few recreational fish- shown in this chapter, the well-being of any recre- eries are managed from an economic perspective, ational fishery is being constantly eroded, not only and this is borne out by the paucity of information by exploitation of the fish directly but also through on the economic value of such fisheries (e.g. degradation of their habitat. Inland waters in Radford 1984; Whelan and Whelan 1986; Kennedy heavily populated areas have to serve the needs of and Crozier 1997; Peirson et al. 2001). Conse- water supply, waste water and sewerage disposal, quently, fisheries are given low priority in any canoeists, sailors, bird watchers, recreational fish- consultation process and it is difficult to attract ers and those who just like to walk along canals, investment or credit for development of a fishery. rivers and lakes. In addition, the demands for sus- To overcome this problem there is an urgent tainable use of the environment that grew out of need to provide accurate social and economic valu- the Rio conference of 1992 have put emphasis on ation of recreational fishing (Aas and Ditton 1998) the need not only to manage exploited resources for justification of its position in conflicts between but also to promote biodiversity. How the conflicts several users. There are various direct methods of interest between these various interests are re- of valuing a fishery, such as financial and cost– solved must depend on involving all stakeholders benefit analysis (e.g. Kennedy and Crozier 1997). in the management process. In Britain, the Envi- Alternative indirect methods such as contin- ronment Agency consults widely over conserva- gency valuation analysis have also been used tion and management plans. In the US similar (e.g. Baker and Pierce 1997; Postle and Moore 1998; stakeholder involvement exists, but in many parts Gautam and Steinback 1998). However, in most of the world there is still little integration between cases these methods rarely give a true economic the different demands placed on waters exploited value to the fishery because they are based on by recreational fishers. Integration and cooperation the willingness of the end-user to pay for the will be essential for their sustainable development. fishing experience and do not account for the value of fishing rights and access rights. If recreational fisheries are to be preserved in the future, it is REFERENCES essential that more appropriate techniques are developed to provide sound economic valuation of Aas, O. and Ditton, R.B. (1998) Human dimensions per- the fisheries. spective on recreational fisheries management. In: P. 386 Chapter 17

Hickley and H. Tompkins (eds) Recreational Fisheries: Contribution of Fisheries to Food Security, Kyoto, Social, Economic and Management Aspects. Oxford: Japan KC/FI/95/TECH/3. Rome: FAO. Fishing News Books, pp. 153–64. Coates, D. (1998) Codes of practice for the stocking and Anon. (1997) BRIEFS, the newsletter of the American introduction of fish. In: I.G. Cowx (ed.) Stocking and Institute of Fishery Research Biologists, 26 (5), 5. Introduction of Fish. Oxford: Fishing News Books, pp. Axford, S. (1991) Some factors affecting angling catches 383–96. in Yorkshire rivers. In: I.G. Cowx (ed.) Catch Effort Cockcroft, A.C., Griffiths, M.H. and Tarr, R.J.Q. Sampling Strategies: their Application in Freshwater (2000) Marine recreational fisheries in South Africa: Fisheries Management. Oxford: Fishing News Books, status and challenges. In: T.J. Pitcher (ed.) Evaluating pp. 143–53. the Benefits of Recreational Fisheries. University Baker, D.L. and Pierce, B.E. (1997) Does fisheries manage- of British Columbia, Fisheries Centre Reports, ment reflect societal values? Contingent valuation Vancouver, vol. 7, no. 2, pp. 64–70. evidence for the River Murray. Fisheries Management Cooke, S.J., Philipp, D.P., Schreer, J.F. and McKinley, R.S. and Ecology 4, 1–16. (2000) Locomotory impairment of nesting largemouth Beaumont, W.R.C., Welton, J.S. and Ladle, M. (1991) bass following catch and release angling. North Comparison of rod catch data with known numbers of American Journal of Fisheries Management 20, 968– Atlantic salmon (Salmo salar) recorded by a resistivity 77. fish counter in a southern chalk stream. In: I.G. Cowx Cowx, I.G. (1991) The use of angler catch data to examine (ed.) Catch Effort Sampling Strategies: their Applica- potential fishery management problems in the lower tion in Freshwater Fisheries Management. Oxford: reaches of the River Trent, England. In: I.G. Cowx (ed.) Fishing News Books, pp. 49–60. Catch Effort Sampling Strategies: their Application in Berg, R. and Rösch, R. (1998) Animal welfare and angling Freshwater Fisheries Management. Oxford: Fishing in Baden-Württemberg, Germany. In: P. Hickley and News Books, pp. 154–65. H. Tompkins (eds) Recreational Fisheries: Social, Cowx, I.G. (1994a) Stocking strategies. Fisheries Man- Economic and Management Aspects. Oxford: Fishing agement and Ecology 1, 15–31. News Books, pp. 88–92. Cowx, I.G. (1994b) Strategic approach to fishery rehabili- Bettoli, P.W. and Osborne, R.S. (1998) Hooking mortality tation. In: I.G. Cowx (ed.) Rehabilitation of Freshwater of striped bass following catch and release angling. Fisheries. Oxford: Fishing News Books, pp. 1–9. North American Journal of Fisheries Management 18, Cowx, I.G. (1995) Western Europe. In: K.T. O’Grady (ed.) 609–15. Review of Inland Fisheries and Aquaculture in the Bunt, D.A. (1991) Use of rod catch effort data to monitor EIFAC Area by Subregion and Subsector. FAO migratory salmonids in Wales. In: I.G. Cowx (ed.) Fisheries Report 509, Suppl. 1, 25–34. Catch Effort Sampling Strategies: their Application Cowx, I.G. (1996) The integration of fish stock assess- in Freshwater Fisheries Management. Oxford: Fishing ment into fisheries management. In: I.G. Cowx (ed.) News Books, pp. 15–32. Stock Assessment in Inland Fisheries. Oxford: Fishing Carlton, F.E. (1975) Optimum sustainable yield as a News Books, pp. 495–506. management concept in recreational fisheries. In: P.M. Cowx, I.G. (1997) L’introduction d’espèces éstrangères Roedel (ed.) Optimum Sustainable Yield as a Concept de poissons dans les eaux douces européennes: in Fisheries Management. American Fisheries Society succès économiques ou désastre écologique? Bulletin Special Publication 9. Bethesda Md.: American Français de la Pêche et de la Pisciculture 344/345, Fisheries Society, pp. 45–9. 57–78. Carvalho, G.R. and Cross, T.F. (1998) Enhancing fish Cowx, I.G. (ed.) (1998a) Stocking and Introduction of production through introductions and stocking: Fish. Oxford: Fishing News Books. genetic perspectives. In: I.G. Cowx (ed.) Stocking and Cowx, I.G. (1998b) Aquatic resource management Introduction of Fish. Oxford: Fishing News Books, pp. planning for resolution of fisheries management 329–37. issues. In: P. Hickley and H. Tompkins (eds) Churchward, A.S. and Hickley, P. (1991) The Atlantic Recreational Fisheries: Social, Economic and Man- salmon fishery in the River Severn (UK). In: I.G. Cowx agement Aspects. Oxford: Fishing News Books, pp. (ed.) Catch Effort Sampling Strategies: their Applica- 97–105. tion in Freshwater Fisheries Management. Oxford: Cowx, I.G. (2001) Recreational sport fishing in fresh Fishing News Books, pp. 1–14. waters. In: P. Safran (ed.) Encyclopedia of Life Support Coates, D. (1995) Inland capture fisheries and enhance- Systems (Theme 5.5 Fisheries and Aquaculture: ment: status, constraints and prospects for food Fish and other Marine and Freshwater Products), security. International Conference of Sustainable Paris: UNESCO. Recreational Fishing 387

Cowx, I.G. and Broughton, N.M. (1986) Changes in the the Third British Freshwater Fisheries Conference, species composition of anglers’ catches in the River University of Liverpool, pp. 89–94. Trent (England) between 1969 and 1984. Journal of Fish Ellis, J.W. (1998) Trends in recreational angling: the Biology 28, 625–36. British waterways experience. In: P. Hickley and Cowx, I.G. and Godkin, P.A. (2000) Analysis of the H. Tompkins (eds) Recreational Fisheries: Social, environmental and economic impact of operations to Economic and Management Aspects. Oxford: Fishing reinforce the aquatic fauna of fresh waters for fishery News Books, pp. 63–9. purposes. Report to the Directorate General XIV, FAO (1996) Precautionary Approach to Capture Fish- European Union. eries and Species Introductions. FAO Fisheries Cowx, I.G. and Welcomme, R.L. (eds) (1998) Rehabilita- Department Technical Guidelines for Responsible tion of Rivers for Fish. Oxford: Fishing News Books. Fisheries, No. 2. Rome: FAO. Cryer, M. and MacLean, G.D. (1991) Catch for effort in FAO (1997) Inland Fisheries. FAO Fisheries Department a New Zealand recreational trout fishery – a model Technical Guidelines for Responsible Fisheries, No. 6, and implications for survey design. In: I.G. Cowx (ed.) Rome: FAO. Catch Effort Sampling Strategies: their Application in Feltham, M.J., Cowx, I.G., Davies, J.M., Harvey, J.P., Freshwater Fisheries Management. Oxford: Fishing Wilson, B.R., Britton, J.R. and Holden, T. (1999) News Books, pp. 61–71. Case studies of the impact of fish-eating birds on Cryer, M. and Edwards, R.W. (1987) The impact of angler inland fisheries in England and Wales. London: ground bait on benthic invertebrates and sediment MAFF/DoE. respiration in a shallow eutrophic reservoir. Environ- Flickinger, S.A., Bulow, F.J. and Willis, D.W. (1999) Small mental Pollution 46, 137–50. impoundments. In: C.C. Kohler and W.A. Hubert (eds) Cryer, M., Corbett, J.J. and Winterbotham, M.D. (1987a) Inland Fisheries Management in North America. The deposition of hazardous litter by anglers at coastal Bethesda, Md.: North American Fisheries Society, pp. and inland fisheries in South Wales. Journal of Envi- 561–88. ronmental Management 25, 125–35. Gardiner, R. (1991) Modelling rod effort trends from Cryer, M., Linley, N.W., Ward, R.M., Stratford, J.O. and catch and abundance data. In: I.G. Cowx (ed.) Catch Randerson, P.F. (1987b) Disturbance of overwintering Effort Sampling Strategies: their Application in wildfowl by anglers at two reservoir sites in South Freshwater Fisheries Management. Oxford: Fishing Wales. Bird Study 34, 191–9. News Books, pp. 92–9. Dekker, W. (1991) Assessment of the historical downfall Gautam, A. and Steinback, S. (1998) Valuation of recre- of the Ijsselmeer fisheries using anonymous inquiries ational fisheries in the north-east US. Striped bas: a for effort data. In: I.G. Cowx (ed.) Catch Effort Sam- case study. In: P. Hickley and H. Tompkins (eds) Recre- pling Strategies: their Application in Freshwater ational Fisheries: Social, Economic and Management Fisheries Management. Oxford: Fishing News Books, Aspects. Oxford: Fishing News Books, pp. 97–105. pp. 233–40. Gautman, A. and Hicks, R. (1999) Using revealed state Department of Fisheries and Oceans, Canada (1998) 1995 and preferences for estimating the benefits of recre- Survey Highlights. Survey of Recreational Fisheries ational fisheries regulations. In: T.J. Pitcher (ed.) in Canada. Ontario: Department of Fisheries and Evaluating the Benefits of Recreational Fisheries. Oceans. University of British Columbia, Fisheries Centre Die, D.J., Piestrepo, V.R. and Hoening, J.M. (1988) Utility Reports, Vol. 7, No. 2, pp. 92–100. per recruit modelling: a neglected concept. Transac- Gerard, P. (1999) Conflict between recreational fishing tions of the American Fisheries Society 117, 274–81. and canoes in a lowland river in Belgium. Fisheries Dill, W.M. (1978) Patterns of change in recreational fish- Management and Ecology 7, 139–44. eries. In: J.S. Alabaster (ed.) Recreational Freshwater Gerard, P. and Timmermans, J.A. (1991) Comparison of Fisheries: their Conservation, Management and fish population estimates and angling catch estimates Development. Medenham, UK: Water Research in a narrow ship canal in Belgium. In: I.G. Cowx (ed.) Centre, pp. 1–22. Catch Effort Sampling Strategies: their Application Edwards, R.W. (1990) The impact of angling on conserva- in Freshwater Fisheries Management. Oxford: Fishing tion and water quality. Proceedings of the Institute of News Books, pp. 127–33. Fisheries Management 21st Annual Study Course, Hickley, P., Marsh, C. and North, R. (1995) Ecological Institute of Fisheries Management, Nottingham, pp. management of angling. In: D.M. Harper and 41–50. A.J.D. Ferguson (eds) The Ecological Basis for Edwards, R.W. and Fouracre, V.A. (1983) Is the banning River Management. Chichester: J. Wiley & Sons, pp. of ground baiting in reservoirs justified? Proceedings of 415–26. 388 Chapter 17

Hickley, P. and Tompkins, H. (eds) (1998) Recreational African line fishery. South African Journal of Marine Fisheries: Social, Economic and Management Science 18, 203–11. Aspects. Oxford: Fishing News Books, pp. 318–30. McKay, H., Furness, R., Russell, I., Parrott, D., Rehfisch, Hodgson, B.P. and Eaton, J.W. (2000) Provison for juvenile M., Watola, G., Packer, J., Armitage, M., Gill, E. and stages of coarse fish in river rehabilitation projects. In: Robertson, P. (1999) The assessment of the effective- I.G. Cowx (ed.) Management and Ecology of River ness of management measures to control damage by Fisheries. Oxford: Fishing News Books. fish-eating birds to inland fisheries in England and Howell, B.R., Moksness, E. and Svåsand, T. (eds) (1999) Wales. York: Central Science Laboratory. Stock Enhancement and Sea Ranching. Oxford: Mike, A. and Cowx, I.G. (1996) The contribution of Fishing New Books, 606 pp. recreational fishing to the fisheries sector in north ICES (1988) Codes of Practice and Manual of Procedures west Trinidad. Fisheries Management and Ecology 3, for Consideration of Introductions and Transfers of 219–28. Marine and Freshwater Organisms. Cooperative Mills, C.A. and Mann, R.H.K. (1985) Environmentally- Research Report, No. 159. Copenhagen: ICES. induced fluctuations in year class strength and their Jefferies, D.J. (1987) The effects of angling interests on implications for management. Journal of Fish Biology otters, with particular reference to disturbance. Insti- 27 (Suppl. A), 209–26. tute of Terrestrial Ecology Symposium 19, 23–30. Morgan, G. (1999) Getting from competition to coopera- Kennedy, G.J.A. and Crozier, W.W. (1997) What is the tion: a choice based decision support system as a value of a wild salmon (Salmo salar L.) smolt? Fisheries consensus tool. In: T.J. Pitcher (ed.) Evaluating the Management and Ecology 4, 103–10. Benefits of Recreational Fisheries. University of Kohler, C.C. and Hubert, W.A. (eds) (1999) Inland British Columbia, Fisheries Centre Reports, Vol. 7, No. Fisheries Management in North America. Bethesda, 2, pp. 163–4. Md.: North American Fisheries Society. Müller, R. (1990) Management practices for lake fisheries Löffler, H. (1991) The recreational fishery in Lake in Switzerland. In: W.L.T. van Densen, B. Steinmetz Constance (Bodensee). In: I.G. Cowx (ed.) Catch Effort and R.H. Hughes (eds) Wageningen: Pudoc, pp. 477–92. Sampling Strategies: their Application in Freshwater Noble, R.L. and Jones, T.W. (1999) Managing fisheries Fisheries Management. Oxford: Fishing News Books, with regulations. In: C.C. Kohler and W.A. Hubert (eds) pp. 108–17. Inland Fisheries Management in North America. Löffler, H. (1998) The pikeperch (Stizostedion Bethesda, Md.: North American Fisheries Society, lucioperca) in Lake Constance: an example of a pp. 455–80. successful introduction? In: I.G. Cowx (ed.) Stocking North, E. (1983) Relationships between stocking and and Introduction of Fish. Oxford: Fishing News Books, angler’s catches in Draycote Water trout fishery. pp. 201–8. Fisheries Management 14, 187–98. Maitland, P.S. (1995) Ecological impact of angling. In: North, E. (2002) Factors affecting the performance of D.M. Harper and A.J.D. Ferguson (eds) The Ecological stillwater coarse fisheries in England and Wales. In: Basis for River Management. Chichester: J. Wiley & I.G. Cowx (ed.) Management and Ecology of Lake and Sons, pp. 443–52. Reservoir Fisheries. Oxford: Fishing News Books. Maitland, P.S. and Lyle, A.A. (1992) Conservation of 284–98. freshwater fish in the British Isles: proposals for O’Hara, K. and Williams, T.R. (1991) Analysis of catches management. Aquatic Conservation: Marine and from the British Angling Championships. In: I.G. Freshwater Ecosystems 2, 165–83. Cowx (ed.) Catch Effort Sampling Strategies: their Malvestuto, S.P. (1983) Sampling the recreational fish- Application in Freshwater Fisheries Management. ery. In: L.A. Nielsen and D.L. Johnson (eds) Fisheries Oxford: Fishing News Books, pp. 214–22. Techniques. Bethesda Md.: American Fisheries Pawson, M.G. (1991) The relationship between catch, Society, pp. 97–419. effort and stock size in put-and-take trout fisheries, its Malvestuto, S.P. (1991) The customization of recreation- variability and application to management. In: I.G. al fishery surveys for managment purposes in the Cowx (ed.) Catch Effort Sampling Strategies: their United States. In: I.G. Cowx (ed.) Catch Effort Application in Freshwater Fisheries Management. Sampling Strategies: their Application in Freshwater Oxford: Fishing News Books, pp. 72–80. Fisheries Management. Oxford: Fishing News Books, Pearce, H.G. (1983) Coarse fish stocking to British rivers pp. 201–14. – case study: the lower Welsh Dee. In: Proceedings of McGrath, M.D., Horner, C.C.M., Brouwer, S.L., the 14th (1983) Institute of Fisheries Management Lamberth, S.J., Mann, B.Q., Sauer, W.H.H. and Eras- Annual Study Course, pp. 209–20. Liverpool: Liver- mus, C. (1997) An economic valuation of the South pool University. Recreational Fishing 389

Peirson, G., Tingley, D., Spurgeon, J. and Radford, A. water Fisheries Management. Oxford: Fishing News (2001) Economic valuation of inland fisheries in Books, pp. 81–90. England and Wales. Fisheries Management and Smith, P.A., Leah, R.T. and Eaton, J.W. (1998) A review of Ecology 8, 414–24. the current knowledge on the introduction, ecology Petts, G.E. (1984) Impounded Rivers: Perspectives for and management of zander, Stizostedion lucioperca, Ecological Management. Chichester: Wiley & Sons. in the UK. In: I.G. Cowx (ed.) Stocking and Int- Pitcher, T.J. (1999) Director’s Foreword. In: T.J. Pitcher roduction of Fish. Oxford: Fishing News Books, pp. (ed.) Evaluating the Benefits of Recreational Fisheries. 209–24. University of British Columbia, Fisheries Centre Steffens, W. and Winkel, M. (1999) Current status and Reports, Vol. 7, No. 2, pp. 4–8. socioeconomic aspects of recreational fisheries in Pollock, K.M., Jones, C.M. and Brown, T.L. (1994) Angler Germany. In: T.J. Pitcher (ed.) Evaluating the Benefits Survey Methods and their Applications in Fisheries of Recreational Fisheries. University of British Colum- Management. American Fisheries Society Special bia, Fisheries Centre Reports, Vol. 7, No. 2, pp. 130–3. Publication, No. 25 Bethesda, Md.: North American Swales, S. and O’Hara, K. (1983) A short term study of the Fisheries Society. effects of a habitat improvement programme on the Postle, M. and Moore, L. (1998) Economic valuation of distribution and abundance of fish stocks in a small recreational fisheries in the UK. In: P. Hickley and H. lowland river in Shropshire. Fisheries Management 14, Tompkins (eds) Recreational Fisheries: Social, Eco- 135–44. nomic and Management Aspects. Oxford: Fishing Templeton, R. (1995) Freshwater Fisheries Manage- News Books, pp. 184–99. ment. Oxford: Fishing News Books. Pottinger, T.G. (1997) Changes in water quality within US Fish and Wildlife Service (1997) 1996 National anglers’ keepnets during confinement of fish. Fisheries Survey of Fishing, Hunting and Wildlife-associated Management and Ecology 4, 341–54. Recreation. National overview. Preliminary findings. Pygott, J.R., O’Hara, K. and Eaton, J.W. (1990) Fish US Department of the Interior, Washington. community structure and management in navigated Walters, C. and Cox, S. (1999) Maintaining quality in British canals. In: W.L.T. van Densen, B. Steinmetz recreational fisheries: how success breeds failure and R.H. Hughes (eds) Management of Freshwater in management of open-access sport fisheries. In: T.J. Fisheries. Wageningen: Pudoc, pp. 547–57. Pitcher (ed.) Evaluating the Benefits of Recreational Raat, A.J.P., Klein Breteler, J.G.P. and Jansen, S.A.W. Fisheries. University of British Columbia, Fisheries (1997) Effects on growth and survival of retention of Centre Reports, Vol. 7, No. 2, pp. 22–9. rod-caught cyprinids in large keepnets. Fisheries Man- Weithman, A.S. (1999) Socioeconomic benefits of agement and Ecology 4, 355–68. fisheries. In: C.C. Kohler and W.A. Hubert (eds) Inland Radford, A.F. (1984) The economics and values of Fisheries Management in North America, 2nd edn. recreational salmon fisheries in England and Wales: Bethesda, Md.: American Fisheries Society, pp. an analysis of the rivers Wye, Mawddach, Tamar and 193–216. Lune. Portsmouth: CEMARE Report, No. 8. Welcomme, R.L. (1988) International Introductions of Roth, E., Toivonen, A.L., Navrud, S., Bengtsson, B., Inland Aquatic Species. FAO Fisheries Technical Gudbergsson, G., Tuunainen, P., Appelblad, H. and Paper, No. 294. Rome: FAO. Weissglas, G. (2001) Methodological, conceptual and Welcomme, R.L. (1992) A history of international sampling practices in surveying recreational fisheries introductions of inland aquatic species. ICES Marine in the Nordic countries – experiences of a valuation Science Symposium 194, 3–14. survey. Fisheries Management and Ecology 8, 355–68. Welcomme, R.L. (1996) Definitions of aquaculture and Russell, I.C., Dare, P.J., Eaton, D.R. and Armstrong, J.D. intensification of production from fisheries. FAO (1996) Assessment of the problem of fish-eating birds Aquaculture Newsletter 12, 3–5. in inland fisheries in England and Wales. Lowestoft: Whelan, B.J. and Whelan, K.F. (1986) The economics of Directorate of Fisheries Research. salmon fishing in the Republic of Ireland, present and Salojärvi, K. and Mutenia, A. (1994) Effects of stocking of potential. In: W.W. Crozier (ed.) Proceedings of the fingerlings on recruitment in the Lake Inari whitefish Institute of Fisheries Management (NI Branch) 17th (Coregonus laveratus L. s.l.) fishery. In: I.G. Cowx (ed.) Annual Study Course. University of Ulster, Coleraine, Rehabilitation of Freshwater Fisheries. Oxford: pp., 191–208. Fishing News Books, pp. 302–13. Wilson, J. (1999) The Complete Book of Fishing: Tackle, Small, I. (1991) Exploring data provided by angling for Techniques, Species, Bait. London: Hamblyn. salmonids in the British Isles. In: I.G. Cowx (ed.) Catch Wolos, A. (1991) Anglers’ opinions as to the quality of Effort Sampling Strategies: their Application in Fresh- fishing and the fishery management in selected Polish 390 Chapter 17

waters. In: I.G. Cowx (ed.) Catch Effort Sampling Wortley, J. (1995) Recreational fisheries. In: K.T. O’Grady Strategies: their Application in Freshwater Fisheries (ed.) Review of Inland Fisheries and Aquaculture in Management. Oxford: Fishing News Books, pp. the EIFAC Area by Subregion and Subsector. FAO 134–42. Fisheries Report 509, Suppl. 1, 60–72. Index

Note: Numbers in italics indicate figures; numbers in bold indicate tables. Cross-references to Volume 1 appear at the end of the relevant entry under see also, and give both volume number and page number. abundance surveys age determination methods, 72 angling litter, 381 extended survivors analysis see also 1.104–7 angling practices, 381–2 (XSA), 152–3 age distribution of population, 72 animal welfare issues, 55 forthcoming year class strength modelling, historical aspects, 73 annual catch, difference equations, estimation, 166 year class fluctuations, 72, 72 119, 120, 120 individual-based models (IBMs), age–length keys (ALK), 279, 280, 281 Anoplopoma fimbrae (sablefish), 243 Alaska fur seal (Callorhinus 319, 320, 326 long-term forecasting, 172 ursinus), 71 Apogon kauderni (Banggai marine protected areas, 294, 295, Alaska Fur Seal Commission, 71 cardinalfish), 328 296, 302–3 Alaska pollock (Theragra Apolemichthys guezi, 322 virtual population analysis (VPA), chalcogramma), 19, 241 Apteryx spp. (Kiwi bird), 254 130, 141, 145; ad hoc tuning, Alaska Seafood Marketing Institute aquaculture 145–6; multispecies model (ASMI), 48 branding, 49 tuning, 158–9 albacore (Thunnus alalunga), 21 consumer environmental see also 1.133, 1.135–6, 1.138, Alca torda (razorbill), 359 concerns, 54, 55, 56 1.194 Allee effect (depensation), 331–2, farmed fish escapes, gene flow access, sociocultural aspects, 94 334 impact, 324–5 Acipenser brevirostrum (shortnose see also 1.6, 1.7, 1.9 marketing environment, 38 sturgeon), 326 Allen, Kenneth, 76 markets, 39, 40, 41, 53 Acipenser sturio (common Ambon damselfish (Pomacentrus product quality, 46 sturgeon), 326 amboinensis), 347 see also 1.2, 1.3, 1.376–7, 1.382–3 see also 1.99, 1.206, 1.207 American buffalo (Bison bison), 254 archaeological records, 63 Acipenseridae (sturgeons), 199, 330 American Fisheries Society, 333 Arctic shrimp (Pandalus borealis), susceptibility to overfishing, 330, Americas, archaeological records, 63 18 331 Ammodytes spp. (sandeels), 18 Arctic skua (Stercorarius AD Model Builder, 276 bird predation, 347, 348, 349 parasiticus), 349 adaptive management, 287 North Sea feeding study, 156, 159 Arctic tern (Sterna paradisea), 348, adaptive traits modelling, 230, predator removal response, 345 349 233–6 Amsterdam albatross (Diomedia Ardea sp. (heron), 378 adaptation approach, 234–6 amsterdamensis), 359 area closures comparative aspects, 236 anadromous migration historical aspects, 71 local interactions, 236 nutrient cycling to terrestrial see also marine protected areas optimization, 234 systems, 349 (MPAs) rule-based approaches, 233–4 see also 1.136, 1.138, 1.176, 1.177, area-specific mortality, 282 strategy vector incorporation, 1.187–8, 1.189, 1.190, 1.191, artificial neural network (ANN) 235, 235–6 1.208, 1.208, 1.209, 1.211, models, 234, 235, 236, 241, 243, advertising, 48 1.240, 1.306 285 African Great Lakes, 368 Anampses viridis (green wrasse), artisanal fisheries, 25–9 see also 1.51, 1.52 322 capture techniques, 13, 25, 26 392 Index artisanal fisheries (cont.) beat fishing, 30 Ecopath model, 215, 216 coral reef, 28 behavioural adaptive traits forecasting, 127–8, 164–87; definition, 25 evolved behavioural strategies, biological reference points, 181; freshwater, 28 234 short-term, 164, 165, 166; total management, 29, 30–1 individual-based models (IBMs), yield curves, 179–81; misconceptions, 29 230, 233–6; adaptation, 234–6; yield/biomass-per-recruit optimal exploitation pattern, 30 comparison of modelling calculations, 173–5 research problems, 25–8 approaches, 236; local mathematical models, 280–1 selectivity, 30 interactions, 236; optimization, Odum’s ecosystem maturity species extirpation, 327 234; rule-based approaches, theory, 220 vessels, 25 233–4; strategy vector surplus production models, 107, Atlantic cod see Gadus morhua incorporation, 235, 235–6 107–8 Atlantic halibut see Hippoglossus planktivore spatial life-history classical differential equations, hippoglossus model, 235, 235–6 109, 110, 111–13; comparisons Atlantic herring see Clupea reinforcement learning, 234 with data, 122; difference harengus Benguela upwelling, 345 equations, 116, 118, 120, 121, Atlantic salmon (Salmo salar), 324, food web, 346; effects of fur seal 121; equilibrium, 112, 113, 113, 330 cull, 354 113 aquaculture markets, 40 benthic carrion feeders, 352–3 trophic levels concept, 217 see also 1.217, 1.382 benthic communities bird scarers, 23 auctions, 43–5 energy subsidies from bottom birds fish quality inspection, 44–5 trawling activities, 352–3 control in recreational fisheries, remote bidding schemes, 45 impact of fishing, 354–8, 356, 357; 378 squential selling strategy, 44 meta-analysis, 357–8, 358 habitat damage through angling Australia, 296, 328, 360 Bering Sea, 214, 345 practices, 381 Australian Society for Fish Biology, Bermuda, 329 see also sea birds 333 Beverton, Raymond, 73, 75 Birdseye, Clarence, 65 automated haulers, 25 Beverton–Holt model, 75, 76, 118, Bison bison (American buffalo), 254 Azerbaijan, 326 127, 178, 179, 205, 272, 276, black-legged kittiwake (Rissa Azurina eupalama (Galapagos 278, 283, 287 tridactyla), 348, 349 damselfish), 322 marine protected areas (MPAs), blast fishing, 329, 335 301, 302 bleeding at slaughter, 46 bag limits, 373, 379 see also 1.125, 1.128, 1.129 blue whale (Balaenoptera Bahaba taipingensis (giant yellow bias, data sampling/stratification, musculus), 69, 76, 77, 79 croaker), 330 89 blue whiting (Micromesistius bait locating behaviour, 22 biblical accounts, 61–2 poutassou), 19 bait/baiting Bigelow, Henry, 63, 64, 73 see also 1.106 gill-netting, 24 bigeye (Thunnus obesus) bluefin tuna (Thunnus thynnus), long lining, 21, 22 long lining, 21, 22, 23 330 machines, 21 see also 1.214, 1.215 long lining, 21, 23, 204 recreational fishing, 368 biodiversity, 258 see also 1.182–3, 1.189 Balaenoptera borealis (sei whale), conservation, 7, 8; marine bluehead wrasse (Thalassoma 79 protected areas (MPAs), 305, bifasciatum), 347 Balaenoptera musculus (blue 309, 310; partitioning systems, boat licences, 264, 265 whale), 69, 76, 77, 79 2 transferable, 256 Balaenoptera physalus (fin whale), safeguards, 334–6 Bocaccio (Sebastes paucispinus), 69, 79 bioeconomics, historical aspects, 327 bald eagle (Haliaeetus 78–9 body size leucocephalus), 349 biological data collection, 92–3 community structure changes Balistidae (triggerfish), 306, 344 biological reference points, 86, 335 due to fishing, 342–3 see also 1.251 dynamic pool long-term forecasts, limits, historical aspects, 70, 71 Banggai cardinalfish (Apogon 175, 181–3 marine protected areas, 294, 296, kauderni), 328 fishing mortality, 181–2 297, 303 Bangladesh, 28 minimum spawning stock size, size–based assessments, 189–208, Baranov, Fëdor, 73, 74 182 279–80 barndoor skate (Dipturus laevis), stock collapse prediction, 182 weight–length relationship, 279 328 biomass see also 1.230–1, 1.258, 1.274, beach seining, 85 accumulation in marine protected 1.275–7, 1.276, 1.286, 1.292, beam trawls, 18, 352 areas (MPAs), 303, 304–5, 305, 1.321–2, 1.326–30, 1.329, historical aspects, 64 307, 307 1.334–6, 1.335, 1.365, 1.375 Index 393

Bolbometopon muricatum California Fish and Game (CFG) fishing models, 278 (bumphead parrotfish), 327 Fisheries Laboratory, 4, 5 virtual population analysis (VPA), bottom long lining, 20 Californian sardine 129, 130 species, 22 (Sardinops caerulea), 70 catch forecasting, 127–8, 164–87 bottom trawling, 14, 15, 18–19, 352 Californian sardine short-term, 128, 164 bag mesh size regulations, 19 (Sardinops sagax), 4–5 spreadsheets, 167, 168 beam trawls, 18 Callorhinus ursinus see also dynamic pool long-term bottom disturbance: habitat (Alaska fur seal), 71 forecasts; dynamic pool degradation, 19, 335, 354–6; Canada, 71, 368, 379 short-term forecasts scavenging fish attraction, 352 overfishing, 327 catch per unit effort (CPUE), 251, bottom types, 18 Canadian Atlantic Fisheries 252 depths, 18 Advisory Committee cost per unit effort relationship, energy subsidies to marine (CAFSAC), 5 253 benthic communities, 352–3 Canadian Department of Fisheries data collection, 92 otter trawl, 18–19 and Oceans (DFO), 5 dynamic pool models, 127 rigid sorting grids, 19 cannibalism, 345, 378 extended survivors analysis selectivity, 19 individual-based models (IBMs), (XSA), 152–3, 154 boxing catch, physical 239, 239 marine protected areas (MPAs), damage/overfilling, 46 see also 1.278, 1.334, 1.336, 300, 303 Brachionichthys hirsutus 1.336–7 optimal exploitation, 255–6 (spotted handfish), 329 canning, 47 recreational fishing assessment, branding, 47, 48–9, 50 historical aspects, 65 375 Brazil, 368 canoe fisheries, 27 stock size relationship, 251, 252, Brevoortia patronus (menhaden), 16 canoe modifications, 27–8 279 Brody growth model, 115, 116 Cape coast, 30 stock stabilization policies, 260, Bronze Age fishing activities, 63 capelin (Mallotus villosus), 16, 19, 261 brook trout (Salvelinus fontinalis) 184, 236, 261, 301 surplus production models, 110; recreational fishing, 372, 377 distribution modelling, 241, comparisons with data, 122 see also 1.29, 1.126–8, 1.136, 242–3 virtual population analysis (VPA), 1.151–2, 1.154, 1.155, 1.156, see also 1.160 130; multispecies models 1.290 capture fisheries marketing tuning, 158–9; tuning, 141–9, Brosme brosme (tusk), 21, 24 environment, 38 144, 146, 148 brown trout (Salmo trutta), 345 capture techniques, 13–32 catch rate, surplus production recreational fishing, 372, 378, 381 artisanal fisheries, 25, 26 models, 107, 107–8 see also 1.29, 1.87, 1.132, 1.134, historical evolution, 63–4 classical differential equations, 1.135, 1.136, 1.157, 1.158, management information 109, 110; equilibrium, 112, 113, 1.161, 1.216, 1.268, 1.289, requirements, 85, 91–2 113, 113 1.324, 1.325 product quality, 45–6 comparisons with data, 122 Brundtland Commission, 7 size selectivity of gears, 14–15 difference equations, 119, 120, Buccinium undatum (whelk), 208 Carangidae (jacks), 304, 343 120, 121 bumphead parrotfish see also 1.56, 1.58, 1.75 catch-and-remove strategy, 372 (Bolbometopon muricatum), Caranx melampygus (trevally), 298 catch-and-return strategy, 372, 327 Caribbean, 294, 296, 310, 327, 330, 379–80, 382 bycatch, 53, 358–60 344 animal welfare, 382 discarding, ecosystem energy carp (Cyprinus carpio) see also 1.241 subsidies, 349–50 recreational fishing, 370, 372, 377, catch-at-age models, 127, 278–9, endangered species, 327 382 285, 286 extinction risk, 328 see also 1.29, 1.50, 1.76, 1.218 see also virtual population fishing methods, 358–9; carrying capacity analysis (VPA) gill-netting, 24, 327 surplus production models, 105, catchability mammals, 360 107; classical differential fishing models, 278 reduction, 29; shrimp trawling, 19 equations, 109, 110, 111, 114; mathematical models, 278 reptiles, 359–60 difference equations, 118, 119, virtual population analysis (VPA) sea birds, 359 121 tuning, 141–5, 144; fitting susceptibility to overfishing, 331 yield–biomass relationship, trends, 148 whiting in cod fisheries, 343–4 105–6 catering, 52–3 see also 1.124, 1.130–1 Catharacta skua (great skua), 348, Caesionidae (fusiliers), 304 catch 351, 352 California Fish and Game (CFG) biomass models, 280–1 caviar, 330 Commission, 70, 71 data collection, 91, 97 cellular automata, 236, 242 394 Index

Centre for Environment, Fisheries Code of Conduct for Responsible coral reef, 212 and Aquaculture Science Fisheries, 84, 86 algal–urchin interactions, 344, (CEFAS), 3 coelacanth (Latimeria chalumnae), 347 cetaceans, 348 331 artisanal fisheries, 28 bycatch, 360 see also 1.26 ecosystem impact of fishing, 343, Chapman, Douglas, 76 cohort analysis, 128–30, 285 344, 347 Charles’s categories of uncertainty, length–frequency data see habitat destruction, 329; by 4, 5 length–frequency analysis bottom trawling, 19, 354 see also 1.3 multispecies, 184; North Sea marine protected areas, 294, 296, Cheilinus undulatus (humphead; feeding study, 157 298, 300, 304, 307, 309, 310 maori; Napoleon wrass), 331 predation mortality, 156 overexploitation, 327–8, 331 Chile, 41, 310 cohort slicing, 201–3 sensitivity to seawater warming, Chilean jack mackerel (Trachurus cold chains, 51 329 murphyi), 16 Cole, Lamont, 78 see also 1.123, 1.306, 1.307, 1.308, chill chains, 51 Cololabis saira (Pacific saury), 1.313–14, 1.314, 1.343, 1.344, chilling, 47 21 1.345, 1.346–7, 1.351–2 China, 40, 50, 321 comanagement, 97, 309–10, 310 cormorant (Phalacrocorax sp.), 378 archaeological records, 63 Committee on the Status of Coryphaena, 198 chinook salmon (Oncorhynchus Endangered Wildlife in Canada Côte d’Ivoire, 27 tshawytscha) (COSEWIC), 333 crab claws removal, 55 closed seasons, 379 common skate (Dipturus batis; croaker (Johnius berlangerii), 207 see also 1.167, 1.187, 1.193 Raja batis), 328, 343 Cushing model, 178, 179 Chromis pelloura (damselfish), 329 common sole see Solea solea cyclic dominance, 78 cichlids common sturgeon (Acipenser Cygnus olor (mute swan), 381 Lake Victoria extinctions, 325 sturio), 326 Cynoscion nebulosus see also 1.50, 1.51–2, 1.53, 1.87, see also 1.99, 1.206, 1.207 (spotted seatrout), 300 1.162, 1.230, 1.234, 1.237 community dependence, data Cyprinus carpio (carp) circle hook, 20 collection, 94 recreational fishing, 370, 372, 377, Cleghorn, John, 65 community structure, 306 382 climate change, 329 data collection, 92 see also 1.29, 1.50, 1.76, 1.218 closed areas, 14, 379 response to fishing, 342, 343; Cyprus, 301, 304 see also 1.194 benthic energy subsidies from closed seasons, 379, 381 carrion, 352–3; temperate dams, 330, 384 Clupea harengus (Atlantic herring) marine fisheries, 344–5; tropical damselfish (Chromis pelloura), 329 age distribution of population, 72, reef, 344 D’Ancona, U., 74, 79, 213 72, 73 see also 1.8–9, 1.321–37, 1.335, data buoys, 97 bird predation, 347, 348 1.341–53 data collection, 84–102, 261 coordinated behaviour modelling, community-based management, 29, biological information, 92 242 91 classification (coding), 88 fishing depletion: distribution competition among fishers, 4 complete enumeration, 89 following, 251; overfishing, compliance information continuous appraisal, 90 254, 330; prey response, 345; dissemination, 92 data quality, 89–90 stock collapse, 176, 326 conservation, 319–36 data sampling, 89–90 marine protected areas (MPAs), listing criteria for commercially economic constraints, 89 301 exploited fishes, 332–3 economic/financial information, North Sea feeding study, 156, objectives, 4 93 160 consumers environmental information, 93 pelagic trawling, 19 environmental awareness, 54 feedback to stakeholders, 95, promotional campaigns, 49 preferences, 97 96–7 purse seining, 16 consumption, data collection, fishery/operations information, stock size estimation, 92 94 91–2 sustainable exploitation, 55 containers, 46 overexploitation risk reduction, see also 1.8, 1.9, 1.73, 1.76, 1.125, contract sales, 43, 45 88 1.160, 1.176, 1.180, 1.190, Convention on International Trade performance indicators, 86–7 1.200, 1.240, 1.275, 1.303 in Endangered Species (CITES), planning, 86, 88, 97, 98–9 clupeoid bird predation, 347, 348, 326, 331, 333 practical aspects, 94–9; 349 coordinated behaviour modelling, data management, 98; cod 242 methods, 96–8; operational Atlantic see Gadus morhua Coordinating Working Party on considerations, 96; Pacific see Gadus macrocephalus Fisheries Statistics, 88 reporting methods, 98 Index 395

precision/accuracy, 88–9 ecosystem energy subsidies, dynamic pool multispecies sample stratification, 89 349–50; scavenging sea birds, forecasts, 167, 170–1, 184–6, sample surveys, 89 350, 350–2, 351, 353 185 sociocultural information, high-grading practices, 350 dynamic pool short-term forecasts, 93–4 management methods 164–71 standards, 88 development, 71 biomass (stock biomass) strategies, 95 mathematical models, 275, 280–1, estimates, 164, 165, 166, stratification, 89–90, 95; 286 171 major/minor strata, 95–6, 96 Dissostichus eleginoides complications, 167–71; changing data management, 98 (Patagonian toothfish), 21, 359 exploitation patterns, 167, 171; presentation, 99–101; graphics, dolphin-friendly tuna, 54 discards, 167, 169–70, 171; 100, 100–1; tables, 99, 99–100 communication campaigns, 49 longer time horizons, 167, DEFRA, 42 dolphins 168–9; multiple fleets, 167, 170; demersal long lining gill-netting bycatch, 24 multiple option calculations, bottom longline, 20–1 tuna fisheries bycatch, 360; purse 167, 168; multiple species, 167, depth, 21 seining, 17–18 170–1, 184 demographic data collection, 94 dredges, artisanal fisheries, 25 current population size estimates, demographic matrix models, 335, drift-net bycatches, 358 164, 165 336 mammals, 29 dedicated computer programs, depensation (‘Allee effect’), 331–2, sea birds, 359 171 334 drifting (pelagic) longlines, 20, 21 exploitation pattern estimates, see also 1.125 drum (Sciaenops spp.), 300 164, 165 deterministic models, 271–2 drying, 47 fishing mortality, 168–9; future detritus recycling, 220 Dutch auction, 44 level assumptions, 164, 165; developing countries, marine dynamic mathematical models, 272 recent exploitation pattern protected areas (MPAs), 307, dynamic pool long-term forecasts, estimates, 164, 165–6 308, 309 172–84 forthcoming year class strength development of fisheries biological reference points, 175, estimation, 166; economic versus personal 181–3 methods/precautions, 166–7 freedoms approach, 6–7 minimum spawning stock size, implementation: management sustainability, 7–8 182–3 options, 168; results, 167, 169; Dicentrarchus labrax (sea bass) multiple species models, 184–6, spreadsheet, 167, 168 recreational fishing, 369 185 method, 165–7 see also 1.185, 1.278 precautions, 173, 183, 186 newly recruiting year class size diesel-powered vessels, historical presentation, 173 estimates, 164, 165; aspects, 64 recruitment estimates, 172, 173 estimation methods, 165 difference equations, 116, 116–18, stochastic stock–recruitment 117 relationships, 183 East Africa, 344 equilibirum properties, 118, stock–recruitment relationship Eastfish, 42, 48 118–22, 120 incorporation, 175–81; fitting echo location equipment population dynamics, 120, 121, relationships, 177–9, 179; artisanal fisheries, 25 277 Maximum Sustainable Yield historical aspects, 64 dikes, 330 (MSY) estimation, 180–1; purse seining, 16 Diomedea exulans (wandering possibility of stock collapse, economic aspects, 6, 249–67 albatross), 23, 359 176–7, 180; total yield curves, data collection, 93 Diomedia albatrus (short-tailed 179–81, 180 fisheries management, 263–5; albatross), 359 yield/biomass-per-recruit effort controls, 264; fisheries Diomedia amsterdamensis calculations, 173–5; discrete policy, 86; individual (Amsterdam albatross), 359 sum-of-products method, 174; transferable quotas (ITQs), Dipturus batis (common skate), pseudocode, 174–5; results, 175, 264–5; information 328, 343 176, 177, 185–6; spreadsheet requirements, 85; Dipturus laevis (barndoor skate), implementation, 175, 175 international issues, 265–7 328 dynamic pool models, 276, 278 fishing effort–fish yield direct marketing, 48 concepts, 127 relationship, 250–2 direct observation, data collection, forecasting processes, 127–8, fishing models, 278 97–8 164–87, 286 fishing rights schemes, 256 discarded catches, 53–4, 170 retrospective data analysis, 127–8 fluctuations in stock abundance: data collection, 91 virtual population analysis see catch stabilization, 260–1; dynamic pool short-term virtual population analysis optimum fleet capacity, 261–3, forecasting, 167, 169–70 (VPA) 262; taxation responses, 263–4 396 Index economic aspects (cont.) 349–50; marine benthic Engraulis ringens (Peruvian Graham’s Law, 69 communities, 352–3; anchoveta), 319, 320, 326 historical aspects, 249 scavenging sea birds, 350, susceptibility to overfishing, open access, 252, 253, 253, 254 350–2, 351, 353; 330 optimal exploitation, 254–7, terrestrial systems, 349 see also 1.2, 1.200, 1.240 258–9, 263; fishing rent fishing impact, 342–61; environmental impact (resource rent), 256, 257, 263; freshwater systems, 345; data collection, 93 maximization of profit, 255, piscivorous sea birds, 347–9; fisheries policy, 85 256–7; volume-dependent removal of predators, 353–4, information requirements, 85 price, 256–7 378; study methods, 347; target Epinephelus striatus recreational activities, 375–6, 385; species harvesting, 342–7; (Nassau grouper), 330 in marine protected areas temperate marine fisheries, Equus caballus (horse), 254 (MPAs), 306–7 344–5; tropical reef, 344 Esox lucius (pike) surplus production models, particle-size spectra, 221–2 recreational fishing, 370 249–50 structure: Odum’s ecosystem see also 1.166, 1.255, 1.269, 1.274, sustainable yield, 252, 252–4; maturity theory, 220, 220; 1.286, 1.287, 1.322, 1.323 bioeconomic equilibria, 253, Ulanowicz’s theory of estuarine ecosystems, 212 253–4, 254 ascendancy, 220–1 habitat destruction, 329 time discounting, 257–60, 259, Ectopistes migratorius oyster activities, 354–5 266–7 (passenger pigeon), 332 Euphausia superba (krill), 353–4 value of fishing effort, 254, 255; Ecuador, 41 European Union, 265 maximization goals, 255–6 eelpout (Zoarces), 300 market for fish, 40, 41; Ecopath, 206, 214, 215–19, 286 effort rate information sources, 42 fisheries management classical differential equations, promotional activities, 49 applications, 224 109, 110; equilibirum, 112, 113, Single Market, 40 flow charts, 217, 218 113 Eurostat, 42 Odum’s ecosystem maturity surplus production models, 107, eutrophication, 328, 384 theory testing, 220 107; comparisons with data, evolved behavioural strategies, particle-size spectra generation, 122, 123 individual-based models (IBMs), 221–2, 222 egg survival 234 spatial distribution approaches, individual-based models (IBMs), Exclusive Economic Zone (EEZ), 223–4 238–9 265 trophic level estimates, 219 measurement, 72 Exocoetidae (flying fishes), 16 Ulanowicz’s theory of ascendancy El Niño Southern Oscillation, 93, see also 1.34, 1.240–1 testing, 220–1 322, 326 exotic species introductions see EcoRanger, 217, 220 see also 1.1 species introductions Ecoseed, 224 elasmobranchs explosive harpoon, 69 ECOSIM, 237, 220, 222–3 bycatch, 331 Extended Survivors Analysis (XSA), fisheries management extinction risk, 306 149–56 applications, 224 see also 1.21, 1.24, 1.72, 1.73, practical aspects, 153–4; Ecospace, 223, 224 1.80, 1.81, 1.106–7, 1.267, pseudo-code, 153 fisheries management 1.348 results, 154, 154, 155 applications, 224 elasticity analysis, 335, 336 Survivors method, 149, 150–2 ecosystem models, 211–25, 286 ELEFAN (electronic length- extinction/extinction risk, 321–5 definitions, 212 frequency analysis), 195–6, 197, causes, 324, 324–5 historical aspects, 80, 211 199 conservation role of marine multispecies models, 213–14 emergent fitness, individual-based protected areas (MPAs), 306 particle-size spectra generation, models, 235 habitat destruction, 324, 325, 326, 221–2, 222 emperors (Lethrinidae), 343, 344 327 spatial considerations, 223–4, employment overfishing, 319, 320, 325–9; 232–3, 233 data collection, 91, 93 economic aspects, 254, 259; spatial detail, 235, 235–6 optimization approach from nontargeted species (bycatch), top-down versus bottom-up ecosystem modelling, 224 328; targeted species, 326–8 control, 222–3 employment policy, 91 palaeoextinctions, 322 ecosystem-based management, 90, endangered species, 326–8 recent, 321–2, 323 224–5, 293 artisanal fisheries, 327 risks to exploited species: see also 1.301 target fishery species, 326–8; modelling, 108, 333–4; with ecosystems listing criteria, 332–3 sustainable utilization, 332, fisheries-related energy subsidies, English Board of Trade fisheries 333 349–53; discarding practices, statistics, 66–7 sea bird bycatches, 359 Index 397

species introductions, 324, 325, fishing effort fishing vessels, 13, 14 326 alternative uses artisanal canoe modifications, see also endangered species (opportunity cost), 254–5 27–8 EZ baiter hook, 20 control, 264; taxation, 263, 264 data collection, 88, 92; methods, data collection, 92 97–8 factory ships, historical aspects, definition, 250, 251 deck layout, 15 69–70 fish yield relationship, 250–2; fishing modelling, 277 Faeroe Islands, 41 fishing mortality, 250–1; stock historical evolution, 63 fecundity distribution differences, 251, investment in new boats, 253, marine protected areas, 296, 298 252 263, 265 recovery following population fishing models, 278, 279, 285 licences, 264 decline, 333 individual-based models (IBMs), optimum capacity, economic see also 1.5, 1.97, 1.222 243 aspects, 261–3, 262 Federation of European Aquaculture measurement methods, 250 tracking, 356 Producers (FEAP), 43 measurement problems, 279 fixed tonnage quotas, 265 fermentation, 65 sustainable yield: economic flatfishes Fiji, 327, 331 aspects, 252, 252–4, 263; beam trawls, 18, 352 fin whale (Balaenoptera physalus), optimal exploitation, 255, 256 benthic carrion feeding, 353 69, 79 value produced, 254, 255; see also 1.207, 1.285 financial data collection, 93 maximization goals, 255–6 flathead mullet (Mugil cephalus), First World War fishing reduction see also catch per unit effort 300 (‘Great Fishing Experiment’), (CPUE) flocking behaviour simulation, 242 68, 68, 75, 79, 301 fishing mortality flood plain artisanal fisheries, 28, 30 FiSAT, 195, 196, 204, 214 dynamic pool models, 127; longer fluctuations in stock abundance, fish pumps, 46 time horizons, 168–9; short- 72–3, 333 FishBase, 216 term forecasting, 164, 165–6 economic aspects, 260–1; Fisheries Board of Scotland, 66 Ecopath modelling, 222 optimum fleet capacity, 261–3, fisheries management, 1 estimation methods, 165–6 262; taxation responses, 263–4 adaptive management approach, Extended Survivors Analysis flying fishes (Exocoetidae), 16 287 (XSA), 154 see also 1.34, 1.240–1 artisanal fisheries, 29–31 extinction risks, 108, 254 focus groups, 97 communication campaigns, 49 fishing effort relationship, 250–1; Fonds d’Intervention et economic aspects: data collection, see also catchability d’Organisation des Marches des 93; international issues, 265–7; long-term forecasting: biological Produits de la Pêche Maritime methods, 263–5 reference points, 175, 181–2; (FIOM), 48 ecosystem-based, 90, 224–5, Maximum Sustainable Yield Food and Agriculture Organization 293 (MSY) estimation, 180–1; stock (FAO), 27, 42, 48, 80 fisher’s perspective, 4 collapse occurrence, 180, 182; Code for Responsible Fisheries, 55 fishery scientist’s perspective, 4 yield/biomass-per-recruit food chains, 217 high seas stocks, 266, 267 analysis, 175 see also 1.302, 1.302, 1.303–5, historical aspects, 69–71; mathematical modelling, 277–8, 1.304, 1.310, 1.311, 1.312 institutions, 69–70; methods, 279 food security, 86 70–1 optimization approach from food supply, data collection, 93 information requirements, 85 ecosystem modelling, 224 food webs, 306 knowledge system requirements, surplus production models, 107, Benguela upwelling, 346 84, 85 107–8; classical differential fishing-related changes, 293, 328; life history key points, 335–6 equations, 109, 112, 113, 113; temperate marine fisheries, objectives, 2, 7, 293; biological comparisons with data, 123; 344–5; tropical reef, 344 reference points, 181 difference equations, 116, 118 modelling, 220, 221, 286; precautionary principle, 86, 335 virtual population analysis (VPA), see also trophic levels recreational fishing see 129, 130, 131; judicious see also 1.8, 1.261–2, 1.279, 1.301, recreational fishing averaging method (JAM), 131–3, 1.303, 1.305–16, 1.307, 1.308, fisheries policy, 86 132, 133, 134; separable, 135, 1.309, 1.321, 1.332–3 performance indicators, 91 135–7, 136, 139, 140, 140 Ford–Walford model, 116 sociocultural factors, 94 fishing rent (resource rent), 256, 257, see also 1.103, 1.104 fisheries science 262, 263 forecasting development of institutions, 3 fishing rights schemes catch/biomass, 127–8, 164–87; historical aspects, 66, 71–5 economic aspects, 256 long-term, 272; short-term, origins, 1–2 fishing rent (resource rent) 164–71 fishing days limitation, 264 capitalization, 256 models, 285, 286 398 Index forecasting (cont.) 319, 320, 326, 332; see also 1.9, see also 1.167 see also dynamic pool long-term 1.140, 1.200, 1.212 gill-netting, 23–4, 327 forecasts; dynamic pool virtual population analysis (VPA), artisanal fisheries, 25, 27; short-term forecasts 131, 131, 132, 133, 135; freshwater, 28 fractional trophic levels, 218–19 tuning, 142 bycatch, 327, 358; mammals, 24, France, 41, 52 yield/biomass-per-recruit 360; reptiles, 360; seabirds, 24 Fratercula arctica (puffin), 348, 349 analysis, 175, 176, 177, 185, float line, 24 freezing, 47 185–6 ghost-fishing, 24 historical aspects, 65 see also 1.5, 1.7, 1.8, 1.32, 1.87–9, ground line, 24 French Frigate Shoals, 214 1.101, 1.125, 1.128, 1.128–9, selectivity, 15, 24 French sardine (Sardina pilchardus), 1.132–3, 1.134, 1.135, 1.138, Gilliam’s rule, 234, 240 65 1.140, 1.142, 1.143, 1.158, see also 1.259 freshwater artisanal fisheries, 28 1.160, 1.161, 1.166, 1.167, gilson winches, 14 freshwater ecosystems 1.176, 1.186, 1.189, 1.190, Ginnosar, 62, 63 recreational fishing, 370–2, 371; 1.192–3, 1.202, 1.212–13, 1.236, GLIM (General Linear Model), 276 stocked fisheries, 372–3 1.240, 1.275, 1.305, 1.348, 1.349 global fish production, 40, 84 response to fishing, 345, 346; Galapagos damselfish current decline, 321, 321 predator removal effects, 354, (Azurina eupalama), 322 see also 1.304 378 game fishing, 369–70 Globefish, 42, 48 see also 1.321–37 game theory, 285 Glycymeris glycymeris, 357, 357 freshwater fishes international management Gompertz growth model, 112 extinction, 334; recent, 321–2, policies analysis, 266–7 Gordon, H. Scott, 79 323; risks of sustainable see also 1.239, 1.260–2, 1.261 grading catches, 46, 53–4 utilization, 332 gannets, 349 Graham, Michael, 69, 75 susceptibility to overfishing, 330 Garstang, Walter, 3, 66–7, 68, 69, 75, Graham’s Law of fishing, 69, 70, 73, Fulmarus glacialis (northern 77, 79 75 fulmar), 350–1, 351 garum, 62, 65 great skua (Catharacta skua), 348, Fulton, Thomas, 66 gastropod benthic carrion feeders, 351, 352 functional groups, Ecopath model, 352, 353 greater lizardfish (Saurida tumbil), 216 gear 206, nr 2•• Fund for the Regulation and artisanal fishery regulations, 30, green wrasse (Anampses viridis), Organization of Fish and 31 322 Aquaculture products (FROM), data collection, 88, 92 Greenland, 41, 71 42 fishing models, 277–8, 281 Greenland halibut (Reinhardtius fur seal cull, 354 historical evolution, 63–4 hippoglossoides), 22, 361 fusiliers (Caesionidae), 304 marine habitat damage, 293 grenadier (Macrouridae), 18 recreational fishing, 368; grey snapper (Lutjanus griseus), 300 gadoid bird predation, 347, 348, 349 restrictions, 380 Grimsby, 64, 66, 67 Gadus macrocephalus (Pacific cod) gene flow ground water abstraction, 384 long lining, 21 farmed fish escapes, 325 groundbaiting, 381 see also 1.278 see also 1.193, 1.206, 1.209 groupers (Serranidae), 298, 322, 327 Gadus morhua (Atlantic cod), 41, genetic algorithm (GA), 234, 235, susceptibility to overfishing, 330 189 240, 241, 285 growth capture techniques, 18, 21, 22, 24 genetically modified fish, 325 estimation from length-frequency codling recreational fishing, 369 Geographical Information Systems data, 190; modal progression distribution following stock (GIS), 261, 283 analysis, 191; see also 1.102–3, depletion, 251 Georges Bank, 345, 355, 361 1.107–9 Extended Survivors Analysis Germany, 41, 52 mathematical models, 114–16, (XSA), 154, 154, 155 Ghana canoe fisheries, 27 115, 276, 279–80 larval drift modelling, 241 ghost-fishing, 24, 360–1 mortality rate relationship, 72; marine protected area (MPA) giant clam (Tridacna gigas), 327 see also 1.140–1 models, 301 giant yellow croaker (Bahaba groynes, 380 North Sea feeding study, 156, 160, taipingensis), 330 gruntfish (Haermulidae), 298 160 gill-nets, 13, 14, 23–4, 194 Gulf of Thailand, Ecopath model, predator–prey relationships, 343; fishing models, 278 217, 218, 221–2, 222 capelin (Mallotus villosus) fishing vessels, 13, 15 Gulland, John, 76 interactions, 184 handling, 13 spawning aggregate susceptibility hanging ratio, 24 habitat choice to overfishing, 330 mesh size, 24 individual-based models, 240, stock collapse, 5, 176, 177, 180, recreational fishing, 368, 370, 373 241, 242 Index 399

see also 1.251, 1.256–62, 1.369, oceanography, 68–9, 80 ideal free distribution, 234 1.374 origins of fishing, 63–5 see also 1.251, 1.256–7, 1.258–9 habitat degradation overfishing, 65–8, 70; ‘Great in situ data collection, 97 bottom trawling, 19, 335, 354–6; Fishing Experiment’ (First income distribution, data see also 1.348 World War impact), 67–8 collection, 94 esturine species, 329 population dynamics, 74–5; individual transferable quotas extinction risk, 324, 327 post-Second World War, 75–8 (ITQs), 264–5 marine benthos, 354–8; ports development, 64–5 fixed tonnage quotas, 265 meta-analysis, 357–8 recreational fishing, 367 share quotas, 265 recovery in marine protected spatial expansion, 63, 64, 64 individual-based models (IBMs), areas (MPAs), 307 statistics (fisheries statistics), 62, 228–44, 285 turtles, 360 66, 67, 70 adaptive traits modelling, 230, haddock see Melanogrammus Hjort, Johan, 69, 71, 72, 74–5, 76 233–6; adaptation approach, aeglefinus hoki (Macruronus novaezelandiae), 234–6; comparison of Haermulidae (gruntfish), 298 18 approaches, 236; local hake (Merluccius), 21 Holt, Ernest, 3, 75, 80 interactions, 236; optimization, Haliaeetus leucocephalus Holt, Sidney, 73, 76 234; rule-based approaches, (bald eagle), 349 Homarus americanus (lobster), 361 233–4; strategy vector halibut home range incorporation, 235, 235–6 Atlantic see Hippoglossus marine protected areas and applications, 237–44; fisheries hippoglossus species, 298, 300 assessment, 243–4 Pacific see Hippoglossus see also 1.349 bookkeeping, 231, 232 stenolepis hook-and-line gear, 28 evaluation, 237 handling practices, product quality, artisanal fisheries, 25 features, 230–6, 231 46 hooks formulation, 236–7 harbour porpoise bycatch, 24 angling litter, 381 habitat choice, 240, 241, 242 Hawaii, 298 archaeology, 63 life-history strategies, 237–8; Hawaiian spiny lobster (Palinurus historical evolution, 63 cannibalistic interactions, 239, marginatus), 361 long line, 20 239; growth/survival in early Hazard Analysis Critical Control size selectivity, 15 stages, 238, 238–9; spawning, Points (HACCP), 46, 51 Hoplostethus atlanticus 239–40 hermit crabs, 353 (orange roughy), 199, 265 mechanistic models, 231, 232 heron (Ardea sp.), 378 bottom trawling, 18 Monte Carlo simulations, 230, herring (Atlantic herring) optimal exploitation, 259 231–2, 234 see Clupea harengus susceptibility to overfishing, 331 responsive environments, 242–3 Herrington, William, 70, 78 see also 1.34, 1.166, 1.194, 1.215, schooling behaviour, 242 high-grading practices, 350 1.240 software implementation, 237 Hippoglossoides platessoides horse (Equus caballus), 254 spatial detail modelling, 232–3, (long rough dab), 343 Howard Johnstone US, 43 233; applications, 240–2; fish Hippoglossus hippoglossus Hume, Graham H.R., 73 distribution, 240–2; (Atlantic halibut), 65 humphead (Cheilinus undulatus), pseudocode, 233; simulation see also 1.36, 1.114 331 experiments, 240 Hippoglossus stenolepis (Pacific Huxley, Thomas Henry, 3, 65, 66, 67 specification of individuals, halibut), 65, 177, 272 see also 1.3 229–30; attribute vector, fishing impact, 343 Hybrid method, 146, 148 229–30; strategy vector, 230 long lining, 21 hybridization super-individuals approach, 230, management regulation, wild/domesticated fishes, 325 242 historical aspects, 70 see also 1.207, 1.216 validation/verification, 237 see also 1.58 hydraulic winches, 64 see also 1.142 historical aspects, 61–81 hydroacoustic equipment Indonesia, 28, 328, 331 age structure modelling, 73 artisanal fisheries, 25 industrialized fishing, 6, 7 ecosystem dynamics, 80; models, purse seining, 16 artisanal fisheries comparison, 26 211 Hydrocynus vittatus (tiger fish), historical aspects, 63–4 fisheries economics, 78–9, 368 techniques, 13 249 hydrography for location of fish Infofish, 42, 48 fisheries management, 69–71 historical aspects, 68–9 Infopeche, 48 fisheries science, 71–5 Kitawara’s Law, 68 Infopesca, 48 gears, 63–4 information, 84–102 marketing, 64–5 i-state distribution models, 229 classification, 88 operational fisheries Iceland, 41 data quality, 89–90 400 Index information (cont.) investment levels, data collection, landing fees, 263 domains, 84–5, 91–4 93, 97 landings, biomass estimations, 281 fisheries data, 91–4 Iran, 326 Lankester, Ray, 65, 67, 74, 77, 79 fisheries information cycle, 86–8, Ireland, 52 Large Marine Ecosystem (LME) 87; analytic framework, 94, 95; Isle of Man, 355, 356 approach, 90 continuous appraisal, 90; data Istiophorus spp. (sailfish), 368, 369, largemouth bass (Micropterus collection strategy, 95; data 370 salmoides), 239, 345, 372, 378 sampling, 89–90; see also 1.8, 1.116, 1.258, 1.270, implementation phase, 86–7, J-hook, 20 1.322, 1.323, 1.326, 1.330, 1.332 99; monitoring system, 87–8; jacks (Carangidae), 304, 343 larval drift planning phase, 98–9 see also 1.56, 1.58, 1.75 individual-based models (IBMs), fisheries management, 85; Japan, 359 241 fisheries policy, 86 archaeological records, 63 modelling spatial detail, 233, 233 knowledge system requirements, fisheries oceanography larval stages 84, 85 development, 68, 80 growth/survival models, 238–9 new data sources, 90 market for fish, 40 marine protected areas, 296, 298, performance indicators, 86, 91 Jasus novaehollaniae (rock lobster), 300 precision/accuracy, 88–9 55 see also 1.98, 1.99, 1.135, 1.136, presentation, 99–101 Johnius berlangerii (croaker), nr 194 1.137, 1.137–40, 1.289–90 standards, 88 Judicious Averaging Method, 166 Lates nilotica (Nile perch), 6, 193, uncertainty, 88 juvenile stages 325, 368 Infosamak, 48 recruitment in marine protected see also 1.279, 1.331 Institut Français de Recherche pour areas, 296, 298 Latimeria chalumnae (coelacanth), L’exploration de la Mer see also 1.98–9, 1.132–3, 1.134, 331 (IFREMER), 42 1.135, 1.135, 1.136, 1.140–1, see also 1.26 Institute of Research and 1.157, 1.179, 1.351 Laurec–Shepherd method, 144–5, Development (IRD), 27 146, 149 Inter-American Tropical Tuna Katsuwonus pelamis lead shot, 381 Commission, 71 (skipjack tuna), 241 League of Nations, 70 International Biological Programme see also 1.72, 1.76 Leiognathus brevirostris (ponyfish), (IBP), 217, 218 Kazakhstan, 326 nr 194 International Center for Living keep-nets, 382 leisure fishing see recreational Aquatic Resources Kenya, 294, 296, 300, 304 fishing Management (ICLARM), 27 killer whale (Orcinus orca), 242 length data, 189, 190, 204–8 International Commission for the kinesis, 234 age distribution of population, 72, Conservation of Atlantic Tunas Kiribati, 41, 331 193; see also 1.105, 1.107 (ICCAT), 42, 71 Kitawara, Tasaku, 68, 77 age–length keys (ALK), 279 International Commission for the Kitawara’s Law, 68 mathematical models, 279; Northwest Atlantic Fisheries, Kiwi bird (Apteryx spp.), 254 biomass estimations, 280; 71 krill (Euphausia superba), 353 discards, 280–1 International Council for the krill surplus hypothesis, 353–4 multispecies yield-per-recruit Exploration of the Sea (ICES), 3, Kyle, H.M., 78, 79 estimation, 204–5, 205 43, 67, 68, 70, 71, 280 offshore migration rate biology programme, 67–8, 80 L25, 15 estimation, 205–6, 206, 206, nr hydrography programme, 67, L50, 15 194 72 L75, 15 trophic level estimation, 206 Multispecies Assessment labelling, 47 weight relationship, 279 Working Group, 156, 157, 159, Lake Malawi, 28 see also 1.102–3, 1.107–9, 1.158 160, 184, 186 Lake Tanganyika, 28 length-frequency analysis, 189–90 international issues see also 1.51, 1.52 ad hoc methods, 194 economic aspects, 265–7 lake trout (Salvelinus spp.), assumptions, 194 game theory analysis of recreational fishing, 372 cohort slicing, 202–3 management policies, 266–7 Lake Victoria, 28, 325 computer packages, 203, 203–4 trade, 39–41 artisanal fishers, 6 continuous recruitment, 207 International Labour Organisation, see also 1.51, 1.52 distributions, 190 88 lakes graphical methods, 195; Modal International Monetary Fund, 88 artisanal fisheries, 28 Progression Analysis (MPA), International Whaling ecosystem structure, 212 195 Commission, 71, 76 Lampetra spp. (lamprey), 194 growth estimation, 190 interviews, data collection, 97 see also 1.75, 1.78 growth rate differences, 194 Index 401

modes, 191, 192, 193, 193; overlap location of fish, historical aspects, marine bycatch, 360; drift-net index calculation, 193–4 68 entanglement, 29; gill-netting, mortality rate estimation, 190 logbooks, 281 24 non-parametric methods, 194, data collection, 97, 98 manatees, 328 195, 196–200; ELEFAN logistic growth model, 112, 249, mangrove, 329 (electronic length–frequency 252, 253, 271, 272, 277–8, 280, maori (Cheilinus undulatus), 331 analysis), 196–7, 198, 200; 287, 301 MAPyA, 42 example, 199–200, 200, 201; London Fisheries Agreement, 75 marginal value theorem, 234 general approach to multiple long line, 20 Marine Biological Association of UK optima, 199, 199; problems, hooks, 20, 21, 22 (MBA), 3 198–9; projection matrix, length, 20 marine ecosystems method, 197–8, 200; SLCA rigging, 20–1 larval drift modelling, 233, 233 (Shepherd’s length-composition storage, 13–14 particle-size spectra, 221–2 analysis), 197, 198, 200 types, 20, 21 Marine Mammal Protection Act parameters to estimate, 190–1 long-line fishing, 13, 14, 20–3 (1972), 360 parametric techniques, 194–5, bait/baiting, 21, 22 marine protected areas (MPAs), 200; MIX, 201, 201, 201–2, 202; bycatch, 358; reptiles, 360; 293–311, 334–5 MULTIFAN, 201 sea birds, 22–3, 359 benefits, 298, 300–5, 308, 308; probability plots, 195 catch distribution, 21 detection, 301–3; fisheries yield rigorous statistical estimation commercial aspects, 21, 23 impact, 300–1, 303; mobile methods, 194 fish bait-locating behaviour, 22 versus sedentary species, 303–4, sample validity, 194 fishing times, 22 304; time-scales, 304–5, 305 statistical analysis, 193 fishing vessels, 13–14, 15 biodiversity conservation, 305, see also 1.105 handling/hauling lines, 13, 22 309, 310 Leslie matrices, 229, 230 historical aspects, 63, 64 biomass accumulation, 303, Lethrinidae (emperors), 343, 344 setting lines, 22 304–5, 305, 307, 307 LFDA (Length Frequency Data size selectivity, 22 comanagement approach, 309–10 Analysis), 204 species, 22 disadvantages/costs, 307, 308 licences for fishing vessels, 264, 265 long rough dab (Hippoglossoides ecosystem modelling, 224; transferrable, 256 platessoides), 343 Ecoseed simulations, 224 life histories Lotka, A., 74, 77, 79 fisheries management community structure changes Lotka–Volterra model, 124, 213, relationship, 309 due to fishing, 343, 343 214, 222, 223, 228 habitat damage recovery, 307 individual-based models (IBMs), Lutjanidae (snappers), 296, 298, 343 implementation, 307–11 229, 230, 233, 237–40; Lutjanus griseus (grey snapper), 300 modelling studies, 301, 304, 305 cannibalistic interactions, 239, Lutra lutra (otter), 381 objectives, 293 239; growth/survival in early prediction of stock recovery, 304 stages, 238, 238–9; spatial McIntosh, William, 66 pulse fishing, 304–5 model for planktivorous fish, mackerel see Scomber scombrus recreational diving, 305, 306, 310 235, 235–6 mackerels (Scombridae), 304 siting, 305, 309 key points for fisheries Macromeris polynyma (surf clam), size range, 293, 294 management, 335–6 208 spillover effects, 298, 299, 303, susceptibility to overfishing, 331 Macruronus novaezelandiae (hoki), 304, 309 see also 1.5, 1.6, 1.97–102, 18 target species effects, 294–8; 1.131–3, 1.149–68, 1.178, Madagascar, 41 abundance, 294, 295, 296, 1.238–9, 1.269–71, 1.278, 1.286, Madeiran sardine (Sardinella 302–3; body size, 294, 296, 297, 1.330, 1.369 maderensis), 27 303; fecundity, 296, 298, 300; lift nets, 25 mahseer (Tor tor), 368, 370 recruitment, 296, 300, 303, 309 light lures Makaira spp. (marlin), 335, 368, 369 tourism, 305, 306, 310 artisanal fisheries, 25 Malaysia, 309 underwater visual surveys, 294, long lining, 21, 22 Maldives, 41 296, 302–3 purse seining, 16 Mallotus villosus (capelin), 16, 19, see also 1.7, 1.279, 1.315 limit/threshold reference points, 86, 184, 236, 261, 301 Marine Stewardship Council (MSC), 335 distribution modelling, 241, 55 limnological information, 93 242–3 mark–recapture studies, 67, 70 line hauler, 13 see also 1.160 marketing, 37–57, 61, 62, 97 ling (Molva molva), 24, 177 mammals communication, 48–50 lingcod (Ophiodon elongatus), 207 culls, 354 definitions, 37 live fish sales, 45–6 habitat damage through angling exchange systems, 43–5 lobster (Homarus americanus), 361 practices, 381 historical aspects, 64–5 402 Index marketing (cont.) measurement units, 88 spatial life-history model for information, 42–3, 53 mechanistic models, 231, 232 planktivorous fish, 235, 235–6 key elements, 38–9, 39 Mediterranean, 294, 296 see also 1.6–7, 1.99, 1.175–95, processing methods, 47–8 Mekong Delta, 28 1.208–10, 1.241, 1.306, 1.333, resource management, 53–4 Melanogrammus aeglefinus 1.349, 1.375 retailing, 51–2 (haddock), 65 Migration Coefficient (MC), 281, stages of process, 37–8 capture techniques, 18, 21 283 markets Georges Bank stock collapse, 326 minimum landing size, 4 catering, 52–3 life history key points, 336 minimum mesh size, 14 definitions, 37 North Sea feeding study, 156 minimum spawning stock size, dynamic aspects, 38 overfishing, 70 long-term forecasts, 182–3 international, 39–41 stock collapse, 332 minnow (Phoxinus phoxinus), 202 wholesale, 51 menhaden (Brevoortia patronus), 16 MIX technique, 201, 201, 201, 202, marlin (Makaira spp.), 335, 368, 369 Merlangius merlangus (whiting) 204 Marshall, Alfred, 78 North Sea feeding study, 156 MIXED MODELS, 276 Marsource, 42 predator–prey relationships, 343 mobile gear, 2 mass-balance models, 215 recreational fishing, 369 Modal Progression Analysis (MPA), particle-size spectra generation, Merluccius (hake), 21 191, 195 221 mesh size Molva molva (ling), 24, 177 match fishing, 369 changes: catch forecasting, 171; Monte Carlo simulations, 207, 217, mathematical models, 106, 124, discard reduction, impact on sea 273 270–88 birds, 351 individual-based models, 231–2 basic concepts, 270–2 fishing models, 278 stochastic dynamic programming complexity, 270 regulations, 75; artisanal (SDP), 234 deterministic, 271–2 fisheries, 30, 31; bottom Morocco, 41 dynamic, 272 trawling, 19; historical aspects, mortality rate fisheries applications, 276–83, 71; mathematical models, 281 estimation from length– 284, 285–6; biomass of Mesolithic fishing activities, 63 frequency data, 190; modal stock/catch/discards/landings, meta-analysis, fishing disturbance progression analysis, 191 280–1; body growth, 279–80; impact on benthic fishing mortality see fishing fishing models, 277–9; fishing communities, 357–8, 358 mortality mortality, 277–8; predation metapopulation models, 232 growth relationship, 72 mortality, 277; seasonal meteorological information, 93 see also 1.158 aspects, 281–3; spatial aspects, Mexico, 327 Mozambique, 41 281–3; stock dynamics, 277; microbial loops, 328 MSFOR, 184, 185 stock–recruitment models, 283; Micromesistius poutassou Mugil cephalus (flathead mullet), value of landings, 280–1 (blue whiting), 19 300 input data, 271, 275 see also 1.106 mullets (Mugilidae), 16 marine protected areas (MPAs), Micropterus salmoides (largemouth Mulloides flavolineatus 301 bass), 239, 345, 372, 378 (yellowstrip goat fish), 298 parameters: estimation, 275–6; see also 1.8, 1.116, 1.258, 1.270, MULTIFAN, 201, 204 selection, 272 1.322, 1.323, 1.326, 1.330, 1.332 MULTIFAN-CL, 207 selection process, 274, 274–5 migration, 38, 71, 72 multiple fleets static, 272 body size relationship, 298 dynamic pool short-term stochastic, 271, 272–4 bottlenecks, susceptibility to forecasting, 167, 170 testing, 270, 286–7 overfishing, 330 models, 272 utility, 270 fishery regulations, 379 multispecies forecasts, 184–6, 286 Mauritania, 41 individual-based models (IBMs), dynamic pool short-term Maximum Economic Yield (MEY), 240–1 forecasting, 167, 170–1 91 international management stock–recruitment effects, 185–6 long-term forecasting, 181 agreements, 265, 267 multispecies models, 213–14, 272, Maximum Sustainable Yield (MSY), marine protected area species, 285 91, 287 300 Ecopath see Ecopath difference equations, 119 mathematical models, 281, 282, mass-balance approach, 215 long-term forecasts, 180–1 282 particle-size spectra generation, marine protected areas (MPAs) nutrient cycling to terrestrial 221–2, 222 impact, 301 systems, 349 simulation models, 214–15; Gulf surplus production models, 106, offshore rate estimation from of Thailand, 215; limitations, 107 length data, 205–6, 206, nr 194 215; North Sea, 214; measurement error, 273, 286 salmon, 77–8 Northeastern Pacific, 214 Index 403

yield-per-recruit analysis, 213–14 North Sea migration-related nutrient see also 1.8 community structure changes cycling, 349 multispecies virtual population due to fishing, 343, 343, 345 recreational fishing, 368 analysis, 156–61, 184, 185, 214, discarded fish consumption by sea sustainable exploitation, 55 277, 285, 286 birds, 350 see also 1.99, 1.124, 1.126, 1.133, cohort analysis, 157–8 Ecopath models, 217 1.136, 1.176, 1.187, 1.190, North Sea multispecies stomach international management 1.217, 1.218, 1.332 sampling study, 156–7, 160 agreements, 265 Oncorhynchus tshawytscha pseudo-code, 158 marine protected areas (MPAs), (chinook salmon) suitability investigation, 160, 300, 304, 307–8 closed seasons, 379 160–1 plaice box, 301, 304, 308 see also 1.167, 1.187, 1.193 tuning, 158–9 protected areas during World open access see also 1.8, 1.273 Wars, 294 economic aspects, 252, 253, 253, multispecies yield-per-recruit sea bird predation, 347–8, 348 254; fleet capacity, 263; total estimation, 204–5, 205 northern fulmar (Fulmarus catch controls, 254 mute swan (Cygnus olor), 381 glacialis), 350–1, 351 optimal exploitation comparison, Norway, 40, 41, 49, 50, 265, 324, 255, 257 15Nstable isotope studies 325, 348 Ophiodon elongatus (lingcod), 207 fractional trophic levels Norway pout, 156, 159 optimal exploitation estimation, 219 nutrient cycling artisanal fisheries, 30 see also 1.306 terrestrial ecosystems from economic aspects, 254–7 Namibia, 27, 41 salmon carcasses, 349 orange roughy see Hoplostethus Napoleon wrass (Cheilinus see also 1.332–3 atlanticus undulatus), 331 Orcinus orca (killer whale), 242 Nash equilibrium, 266 ocean circulation models, 233, Organization for Economic Nassau grouper (Epinephelus 235 Cooperation and Development striatus), 330 oceanographic information, 93 (OECD), 42 National Marine Fisheries Service, historical aspects, 68–9, 80 origins of fishing, 63–5 42 oceanographic variables OSMOSE, 221 negative consumer perceptions, data collection systems, 88 otolith studies, 72, 280 47–8 larval drift modelling, 233, larval dispersal in coral reef, net bin (net storage), 13, 14 233 347 net haulers, 13 planktivorous fish behaviour/life see also 1.105, 1.106, 1.133 artisanal fisheries, 25 history modelling, 235 otter (Lutra lutra), 381 net winch, 14 year class fluctuations, 72–3 otter trawl, 18–19 Netherlands, 53 octopus, 195, 304 historical aspects, 64 nets, 28 odour trail detection, 22 overfishing, 1, 3 historical evolution, 63 Odum’s ecosystem maturity theory, bioeconomics, 78, 79 New Caledonia, 294 220, 220 Californian sardine New Wave, 43 offal scavengers, 350–1 (Sardinops sagax), 4–5 New Zealand, 294, 296 offences, data collection, 92 ecosystem impacts, 328–9 New Zealand grayling (Prototroctes oil for preservation, 65 extinction risk, 325–9; economic oxyrhynchus), 326 omega-3 polyunsaturates, 47, 49 aspects, 254, 259 Newfoundland, 359 omnivory index, 219 Graham’s Law, 69 Newton–Raphson method, 129 onboard processing, historical historical aspects, 65–8, 70; First Nile perch (Lates nilotica), 6, 193, aspects, 64 World War, ‘Great Fishing 325, 368 Oncorhynchus mykiss Experiment’, 68, 68, 75 see also 1.279, 1.331 (rainbow trout) interaction with environmental noctural fishing recreational fishing, 372, 377, 381, factors, 326 artisanal lake fishing, 28 382, 383 population susceptibility, 329–32; purse seining, 16 see also 1.29, 1.76, 1.77, 1.216 bycatch, 331; depensation nomadic seasonal fishing activities, Oncorhynchus nerka (‘Allee effect’), 331–2; highly 94 (sockeye salmon) valuable fish, 330–1; limited Nordmore grid, 19 migration, individual-based physiogeographic range, 330; North Atlantic Fisheries models, 241 low productivity, 331; Organization (NAFO), 42 population oscillations, 77, 77–8 shoaling/migration aggregates, North Pacific, Ecopath models, see also 1.125–6, 1.126, 1.139, 330 217 1.160, 1.273, 1.333 targeted species, 326–8 North Pacific Halibut Commission, Oncorhynchus spp. (Pacific salmon) Overfishing Convention, 75 70, 71 conservation, 330 oyster reefs, 354–5 404 Index

Pacific cod (Gadus macrocephalus) piscivorous fishes estimation, 156, 158–9, 159, long lining, 21 community structure changes 184, 277; see also 1.277 see also 1.278 due to fishing, 343 North Sea multispecies study, Pacific rockfish see Sebastes spp. culling in recreational fisheries, 156, 158–9, 159 Pacific salmon see Oncorhynchus 378 recreational fisheries spp. North Sea multispecies study, management, 378 Pacific saury (Cololabis saira), 21 156, 158–9, 159 sea birds, 347–8, 348 packaging, 47, 52 see also 1.267–80, 1.301, 1.328, predator–prey interactions, 79–80, paddlefishes, endangered status, 326 1.329, 1.330, 1.331–2 285 pair trawling, historical aspects, 63 plaice see Pleuronectes platessa community structure changes Palinurus marginatus plaice box, 275, 301, 304, 308 due to fishing, 342–3, 345 (Hawaiian spiny lobster), 361 planktivores, adaptive multispecies models, 213 Palinurus sp. (rock lobster), 94 behaviour/spatial modelling, North Sea multispecies study, Pandalus borealis (Arctic shrimp), 235, 235–6 156–7, 159 18 platforms, 380 see also 1.7, 1.137, 1.137, 1.260–2, panel surveys, 97 Pleuronectes platessa (plaice) 1.261, 1.272–3, 1.273, 1.274, parrotfishes (Scaridae), 304 beam trawling, 18 1.278, 1.313, 1.313–14, 1.314 see also 1.91, 1.238, 1.251 benthic carrion feeding, 353 predators, ecosystem effects of particle-size spectra, 221–2 migration–body size relationship, removal, 328, 342, 343, 346, partitioning systems, 2 298 353–4 passenger pigeon (Ectopistes North Sea plaice box, 275, 301, freshwater systems, 345, 346 migratorius), 332 304, 308 temperate marine fisheries, 344–5 Patagonian toothfish (Dissostichus overfishing/recovery, 67–8, 68 tropical reef, 344 eleginoides), 21, 359 see also 1.156, 1.166, 1.167, 1.180, see also 1.279, 1.301 Pauly, Daniel, 196 1.182, 1.185–7, 1.186, 1.189, prey Pearl, Raymond, 74 1.190, 1.191, 1.241 ecosystem effects of predator Pearl’s logistic equation, 74, 75 Pollachius virens (saithe), 16 removal, 342 pelagic (drifting) longline, 20, 21 North Sea feeding study, 156 predation losses calculation for pelagic long lining, 21 pollution, 384 catch forecasting, 184 bait, 21 polymodal analysis, 193 see also predator–prey species, 22 Pomacentrus amboinensis interactions pelagic trawling, 19–20 (Ambon damselfish), 347 price of fish fishing depth, 20 ponyfish (Leiognathus brevirostris), data collection, 93 mesh size, 20 nr 194 fishing models, 278 trawl instrumentation, 20 Pope’s algorithm, 139 volume-dependent, 256–7 Penaeus duorarum (pink shrimp), population dynamics primary production, 219 300 historical development, 74–5; Pristis pectinata (smalltooth performance indicators, 86, 91 post-Second World War, 75–8 sawfish), 326, 327 data collection, 86–8 mathematical models, 120, 277, probability plots, 195 monitoring, 87–8 285 PROBUB model, 214 profitability estimation, 93 population oscillations, 77, 77–8 process error, 273, 286 personal selling, 48 surplus production see surplus processing Peru, 41 production models historical aspects, 65 Peruvian anchoveta see Engraulis see also 1.334–6, 1.337 marketing aspects, 47–8 ringens population oscillations, 77–8 seasonal activities, 94 Phalacrocorax sp. (cormorant), 378 see also 1.335 products Philippines, 294, 296, 300, 304, 305, ports, historical development, 64–5 data collection systems, 88 310, 343 Portugal, 41 traceability, 46 Phoxinus phoxinus (minnow), 202 pot fishing, 356 profitability, data collection, 93 Physeter macrocephalus bycatches, 358 projection matrix method, 197–8, (sperm whale), 69 ghost-fishing, 360, 361 200 pike (Esox lucius) prawns, 195 promotional activities, 48, 49 recreational fishing, 370 trawl fisheries bycatch, 328 prosecutions, data collection, 92 see also 1.166, 1.255, 1.269, 1.274, precautionary approach, 86, 88, 293, Prototroctes oxyrhynchus (New 1.286, 1.287, 1.322, 1.323 333, 335, 336 Zealand grayling), 326 pike perch (Stizostedion lucioperca), information uncertainty, 88 publicity, 48 378 surplus production models, 108 puffin (Fratercula arctica), 348, 349 pilchards, stock size estimation, 92 predation purse line, 16 pink shrimp (Penaeus duorarum), mortality: multispecies purse seines, 13, 14, 63 300 interactions, 184–5, 185; rate hanging ratio, 16 Index 405

net mesh size, 16 conflicts of interest, 383, 384 release programmes, 71 size, 16 definitions, 367, 369 remote sensing, 54, 90 purse seining, 15–18, 85 development/maintenance, data collection, 97 artisanal fisheries, 27 380–5; angling practices, 381–2 rents bycatch, 359; dolphins, 360 domestic consumption, 370 data collection, 93 fishing vessels, 14, 15; size-related fishery regulations, 379, 379–80 fishing (resource rent), 256, 257, fishing capacity, 16, 17 freshwater fisheries: catch-and- 262, 263 herring overfishing, 254 remove, 372; catch-and-return, reptile bycatch, 359–60 mechanical hauling, 254 372; natural, 370–2, 371; reservoirs, 384 shoaling fish, 330; location, 16 stocked, 372–3, 381–2 resource ownership, 4, 6 put catch and return fisheries, 373 game fishing, 369–70 resource rent (fishing rent), 256 put, grow and take fisheries, 372 gears, 368 resource management put and take fisheries, 372–3, 382 habitat management, 380, 381 information requirements, 85 closed areas, 379 leisure fishing, 369 marketing, 53–4 management, 376, 376–8, 385; retail marketing, 51–2 quality assessment at auctions, control of fish-eating birds, 378; Ricker model, 118, 121, 178, 179, 44–5 predator control, 378; species 228, 272, 302 quality standards, 49 diversity, 377–8; stock see also 1.124, 1.125, 1.127, 1.128, questionnaires, 97 enhancement, 376–7, 377, 1.129 quotas, 53, 263 382–3; unwanted species Ricker, William, 78, 79 fishing rent (resource rent) elimination, 378 Rissa tridactyla (black-legged capitalization, 256 match fishing, 369 kittiwake), 348, 349 individual transferable (ITQs), origins, 367 rivers, artisanal fisheries, 28 264–5 popularity, 367–8 rock lobster (Jasus novaehollaniae), remote electronic sea fisheries, 373 55 monitoring/recording, 54 specimen/specialist anglers, 370 rock lobster (Palinurus sp.), 94 short-term catch forecasts, 164 subsector interactions, 383, 383–5 rockfishes see Sebastes spp. value, 385 rod and line recreational fishing, 368 rabbitfish (Siganidae), 304 recruitment Ross, R., 74 rainbow trout (Oncorhynchus catch/biomass long-term round sardine (Sardinella aurita), 27 mykiss) forecasting, 172, 173; Russia, 326 recreational fishing, 372, 377, 381, see also 1.140, 1.141 382, 383 depensation (‘Allee effect’), 331–2 sablefish (Anoplopoma fimbrae), see also 1.29, 1.76, 1.77, 1.216 individual-based models (IBMs), 319, 320, 326 Raja batis (common skate; 243 Sahel, 28 Dipturus batis), 328, 343 marine protected areas, 296, 298, sailfish (Istiophorus spp.), 368, 369, Raja radiata (starry ray), 343 300, 301, 302, 309 370 random sampling, 89 mathematical models, 276, 283; saithe (Pollachius virens), 16 random walk, 234, 241 see also 1.141–2 North Sea feeding study, 156 rays, susceptibility to overfishing, North Sea multispecies virtual sales promotion, 48, 49 331 population analysis, 159, 159 Salmo salar (Atlantic salmon), 324, razorbill (Alca torda), 359 surplus production models, 116, 330 recovery 117, 118, 119 aquaculture markets, 40 benthic communities following variability, 78, 79, 183, 238, 309; see also 1.217, 1.382 fishing disturbance, 358 see also 1.6, 1.131, 1.132, Salmo trutta (brown trout), 345 First World War fishing reduction 1.132–3, 1.133, 1.134, 1.135, recreational fishing, 372, 378, 381 (‘Great Fishing Experiment’), 1.135, 1.139–40, 1.154 see also 1.29, 1.87, 1.132, 1.134, 68, 68 Red List of Threatened Species, 322, 1.135, 1.136, 1.157, 1.158, long-term forecasts, 172 326, 327, 331 1.161, 1.216, 1.268, 1.289, surplus production modelling, listing criteria, 332–3 1.324, 1.325 121–2, 122 redfish see Sebastes spp. salmon recreational diving, 305, 306, 310 refrigerated fishing vessels, 14, 19 farming, 48; branding, 49; recreational fishing, 28, 367–85 refuges promotional campaigns, 50 animal welfare issues, 382 Ecosim modelling, 222–3 management methods assessment methods, 373–6, 376; see also 1.289–93, 1.331 development, 71 catch per unit effort (CPUE), registration, data collection, 97 migration, 77–8; see also 1.6, 1.78, 375; economic aspects, 375–6; reinforcement learning, individual- 1.176, 1.187 off-site event recall, 374, 374, based models (IBMs), 234 recreational fishing, 269, 368 375; on-site intercept methods, Reinhardtius hippoglossoides salmonids 374, 374, 375 (Greenland halibut), 22, 361 conservation, 330 406 Index salmonids (cont.) see also 1.8, 1.269 sei whale (Balaenoptera borealis), recreational fishing, 368, 369, 370, Sciaenops spp. (drum), 300 79 377, 381 Scomber scombrus (mackerel) seines susceptibility to overfishing, 330 migration patterns, 38; artisanal fisheries, 25, 30; salt-marsh, 329 see also 1.176, 1.186 freshwater, 28 salting, 47, 62, 65 North Sea feeding study, historical evolution, 63 Salvelinus fontinalis (brook trout) 156 see also purse seines recreational fishing, 372, 377 pelagic trawling, 19 selection factor, 15 see also 1.29, 1.126–8, 1.136, prey response to fishing depletion, selection range, 15 1.151–2, 1.154, 1.155, 1.156, 345 selectivity, 14–15, 29–30 1.290 promotional campaigns, 49 artisanal fisheries, 30 Salvelinus spp. (lake trout), purse seining, 16 bottom trawling, 19 recreational fishing, 372 stock collapse, 176 gill-netting, 24 Samoa, 310 see also 1.1, 1.8, 1.72 long lining, 22 sample stratification, 89 Scombridae (mackerels), 304 trawl bag, 19 sample surveys, 89 Scottish fishing experiment, 66, 67, semipelagic bottom longline, 20, 21 sanctuary areas, 379 79 Senegal, 27, 41 sand lance see Ammodytes spp. scup (Stenotomus chrysops), 71 canoes fisheries, 25, 27; range of sandeels see Ammodytes spp. Scyllarisdes squammosus (slipper operation, 27 Sanopus splendidus (splendid lobster), 361 Serranidae, 298, 343 toadfish), 329 sea bass (Dicentrarchus labrax) extinction risk, 306 Sardina pilchardus (French sardine), recreational fishing, 369 see also groupers 65 see also 1.185, 1.278 set-nets, bycatch/ghost-fishing, 360 sardine fisheries, 4–5, 76, 326 sea basses see Serranidae Sette, Oscar, 73 Sardinella aurita (round sardine), 27 sea birds Seychelles, 41, 306, 331 Sardinella maderensis bycatches, 359; gill-netting, 24; shads, susceptibility to overfishing, (Madeiran sardine), 27 long lining, 22–3 330 sardines fish predation, 347–8, 348 share quotas, 265 pelagic trawling, 19 fishing-related prey removal, sharks promotional campaigns, 49 347–9 finning, 55 Sardinops caerulea population impact of industrial length-frequency analysis, 199 (Californian sardine), 70 fisheries, 348–9; energy susceptibility to overfishing, 331 Sardinops sagax subsidies from scavenging, 350, shelf life, 52 (Californian sardine), 4–5 350–2, 351, 353 Shepherd-type stock-recruitment sashimi, 21, 330 Sea Fish Industry Authority (SFIA), analysis, 178, 179 satellite data, 90, 97 42, 48 see also 1.125 Saurida tumbil (greater lizardfish), sea snake bycatches, 361 Shepherd’s length-composition 206, nr 194 sea urchins, 344 analysis (SLCA), 197, 198, 200 scale rings, 72 benthic carrion feeding, 352 Shetland Islands, 351 see also 1.105–6 Seafood Business, 43 shoals scallop dredging disturbance, 352, seagrass beds, 328, 329 pelagic trawling, 20 352, 355, 356 SeaLab, 240 purse seining, 16 Scandinavian countries, 41 seals, 342, 348 susceptibility to overfishing, 330 Scaridae (parrotfishes), 304 gill-netting bycatch, 24 short-tailed albatross see also 1.91, 1.238, 1.251 searching behaviour (Diomedia albatrus), 359 scavenging individual-based models (IBMs), short-term forecasts, 272 fish, 352 234 methods, 164 invertebrates, 352–3 see also 1.271, 1.288, 1.291 see also dynamic pool short-term sea birds, 350, 350–2, 351, 353 seasonal closures, 379 forecasts Schaefer Curve, 76, 273–4 historical aspects, 70, 71 shortnose sturgeon (Acipenser Schaefer, Milner, 76 seasonal fishing activities, 94 brevirostrum), 326 Schaefer stock–recruitment mathematical models, 281–3 shrimp trawls, 29 relationship, 106, 118, 249 Sebastes paucispinus (Bocaccio), bycatch, 19, 327, 360, 361 marine protected areas (MPAs), 327 Siganidae (rabbitfish), 304 301 Sebastes spp., 19, 199, 300 Siganus canaliculatus (white- schooling population decline, 327 spotted rabbitfish), 203, individual-based models (IBMs), susceptibility to overfishing, 202 236, 242 331 sightings, data collection, 92 purse seining, 16 see also 1.207 size see body size Index 407 size selectivity see selectivity marine protected areas, 296 Stercorarius parasiticus size-based assessments, 189–208, movement out of marine (Arctic skua), 349 279–80 protected areas, 300 Sterna paradisea (Arctic tern), 348, size-frequency analysis, 189–208 spawning stock biomass (SSB) 349 modes, 190–4 marine protected areas (MPAs), Stirling Aquaculture, 43 population age distribution, 301, 302 Stizostedion lucioperca (pike perch), 190–204 mathematical models, 280, 281 378 skipjack tuna (Katsuwonus spearing, 28 stochastic dynamic programming pelamis), 241 species, data collection standards, (SDP), 234, 236, 241 see also 1.72, 1.76 88 stochastic models, 271, 272–4 SLCA (Shepherd’s length- species introductions measurement error, 273 composition analysis), 197, 198, extinctions following, 324 probability distributions, 273, 200 precautionary approach, 86 273, 276 slipper lobster (Scyllarisdes recreational fishing, 377, 378, process error, 273 squammosus), 361 382–3 stock assessment, 3, 127–8 smalltooth sawfish (Pristis see also 1.216, 1.279, 1.331, 1.331, data collection, 91, 92 pectinata), 326, 327 1.376 fishing effort measurement, smoking, 47 specimen/specialist anglers, 370 250–1 snappers (Lutjanidae), 296, 298, sperm whale (Physeter two-phase approach, 5–6, 6 343 macrocephalus), 69 uncertainty, 4, 5 snoods, 21, 22 splendid toadfish (Sanopus stock collapse, 319, 320, 326 sociocultural aspects, 2 splendidus), 329 biological reference points, economic optimal exploitation, sponges, 354 182 255–7 spotted handfish (Brachionichthys long-term catch forecasts, 176–7, employment, 65 hirsutus), 329 180 fish population decline impact, spotted seatrout (Cynoscion stock–recruitment relationship, 319 nebulosus), 300 177, 178, 183 management information, 85, 91; sprats, promotional campaigns, stock enhancement data collection, 93–4 49 data collection, 91 marine protected areas (MPAs), Sprattus sprattus (sprat), 16 recreational fishery management, 307, 308, 309 bird predation, 347, 348 376–7, 377, 382–3 sockeye salmon see Oncorhynchus North Sea feeding study, 156 see also 1.215–17, 1.279 nerka pelagic trawling, 19 stock–recruitment relationship, sole ownership, 257, 259 spreadsheets, 190, 279 172 Solea solea (common sole) squid, 195, 304 incorporation into long-term beam trawling, 18 Sri Lanka, 360 catch forecasts, 175–81; fitting benthic carrion feeding, 353 stable isotope studies relationships, 177–9, 179 see also 1.36, 1.186, 1.348 fractional trophic levels mathematical modelling, 283 sorting grids, 19 estimation, 219 multispecies models, 185–6 South Africa, 294, 296 terrestrial ecosystems, nutrient stochastic relationships, 183 South/Southeast Asia, 40, 368 cycling from salmon carcasses, stock collapse, 177, 178, 183 southern bluefin tuna (Thunnus 349 yield-per-recruit calculations, maccoyii), 319, 320, 330, 333, see also 1.306, 1.307, 1.307–8, 173 359 1.308 see also 1.6, 1.124–5 Soviet states, former, 40, 41, 265 stake nets, 46 Stolephorus sp., 94 spatial modelling standards for data collection, 88 stomach contents analysis fish distribution, 240–2, 281–3, starfishes, 353 bottom scavenging fish, 352 282 starry ray (Raja radiata), 343 North Sea multispecies feeding individual-based models (IBMs), state space models, 271 study, 156, 160, 277 232–3, 233; applications, 240–2; static mathematical models, 272 see also 1.115, 1.278, 1.305–6, pseudocode, 233; simulation Statistical Analysis System (SAS), 1.307, 1.309 experiments, 240 276 storage following capture, 46 multispecies production models, statistics (fisheries statistics) Stronglyocentrotus franciscanus 286 First World War fishing reduction (urchin), 207 spawning (‘Great Fishing Experiment’), structured interviews, 97 aggregates, vulnerability to 68, 68, 75 structured models, 228 overfishing, 330 historical aspects, 62, 66, 67, 70 individual-based, 229, 230 individual-based models (IBMs), steam-powered vessels, 63, 64 sturgeons (Acipenseridae), 199, 239–40 Stenotomus chrysops (scup), 71 330 408 Index sturgeons (Acipenseridae) (cont.) see also 1.105, 1.183–4, 1.185, tiger fish (Hydrocynus vittatus), 368 susceptibility to overfishing, 330, 1.186–7, 1.189, 1.190 time discounting, 257–60, 259, 331 target reference points, 86, 335 266–7 subsidies, data collection, 93 taxis, 234 TNS market information subsurface longlining, 21 Technical Consultation on the consultants, 43 super-individual modelling Precautionary Approach to Tor tor (mahseer), 368, 370 approach, 230, 242–3 Capture Fisheries and Species Total Allowable Catches (TACs), 4, supermarkets, 51, 52 Introductions, 86 14, 31 surf clam (Macromeris polynyma), temperature economic aspects, 260–1, 263 207 data collection, 93 international management surface water abstration, 384 indiviual-based models: agreements, 265 surplus production models, 76, migration, 241; spawning problems, 264 105–25, 181, 249–50, 250, 272, strategies, 239–40 short-term catch forecasts, 164; 276, 286 see also 1.87, 1.89, 1.99–102, management options, 167, classical differential equations, 1.100, 1.110–11, 1.111, 1.112, 168 109, 109–12, 110; equilibirum, 1.114, 1.161, 1.183, 1.184, total catch controls, 254 112–14, 113, 113 1.218, 1.227, 1.293, 1.321, historical aspects, 71 comparisons with data, 122–3 1.323–4, 1.324, 1.348, 1.350 total yield curves, long-term economic aspects, 249–50 temperature control forecasting, 179–81, 180 growth models, 114–16, 115 onboard, 46 pseudo-code, 180 key concepts, 107, 107–9 transfer processes, 46 Totoaba macdonaldi (totoaba), marine protected areas (MPAs), wholesale markets, 51 326–7 303 temporal multispecies production tourism, 258 model extensions, 123–4 models, 286 marine protected areas (MPAs), modern difference equations, 116, terrestrial ecosystems, nutrient 305, 306, 310 116–18, 117; equilibirum cycling from salmon carcasses, visitor-related damage, 310 properties, 118, 118–22 349 Trachurus murphyi (Chilean jack multispecies models, 286 territorial behaviour, 298 mackerel), 16 survey data see abundance surveys tetracycline tracking tracking Survivors analysis, 150–2, 154, 155, larval dispersal in coral reef, fishing vessels, 356 165, 171 347 movement of target species across see also Extended Survivors see also 1.191 MPA boundaries, 298 Analysis (XSA) Thailand, 28, 41 trade volume/value, data collection, sushi, 21, 330 Thalassoma bifasciatum (bluehead 93 sustainable utilization, 4, 7–8, 84 wrasse), 347 traditional fishing implements, 25 consumer environmental Theragra chalcogramma (walleye trammel nets, 24 awareness, 55 pollock; Alaska pollock), 19, transfer processes, product quality, economic aspects, 252, 252–4, 241 46 263; bioeconomic equilibria, Thompson, William F., 70, 73 transferable boat licences, 256 253, 253–4, 254; fundamental Thompson, William R., 74 transport criteria, 254; open access, 252, Thomson and Bell length-based historical aspects, 64–5 253, 253, 254, 263; optimal yield equation, 204–5 temperature-controlled exploitation, 254–7, 256, 263 threatened species environemnt, 51 extinction risk modelling, 333–4 Red List criteria, 332 trapping, 28, 46 information requirements, 85, 88 see also endangered species ghost-fishing, 361 long-term data collection, 90 Thunnus alalunga (albacore), 21 recreational fishing, 368 precautionary approach, 86 Thunnus albacares (yellowfin tuna), trawl warp, 18, 20 Red List rate-of-decline criteria, 195, 241 trawling, 18–20 332–3 long lining, 21 bycatch, 358; extinction risk, 328; surplus production concept, 105, see also 1.213–15, 1.214 reptiles, 360 250 Thunnus maccoyii (southern historical aspects, 63, 64 Swarm, 237 bluefin tuna), 319, 320, 330, multispecies yield-per-recruit sweep winches, 14 333, 359 estimation, 204–5, 205 swimbladder, financial value, 326, Thunnus obesus (bigeye) pelagic, 19–20 330–1 long lining, 21, 22, 23 product quality, 46 swordfish see Xiphias gladius see also 1.214, 1.215 see also bottom trawling Thunnus thynnus (bluefin tuna), trawls, 13, 14, 63 tagging studies 204, 330 bag selectivity, 19 movement of target species across long lining, 21, 23 benthic habitat destruction, 19, MPA boundaries, 298, 300 see also 1.182–3, 1.189 335, 354–6 Index 409

benthic invertebrate community United States, 71 bait location, 22 destruction, 329 marine protected areas, 307; see also 1.270, 1.275, 1.334 bottom, 14, 18 Florida, 296, 300; North Volterra, V., 74, 77, 79, 213, 286 fishing models, 277–8 American East Coast, 294 von Bertalanffy, Ludwig, 75 fishing vessels, 14, 15 market for fish, 40 von Bertalanffy model, 111, 112, pelagic, 20 overfishing, 327 114–15, 115, 116, 189, 191, 195, size selectivity, 14–15 US Bureau of Fisheries Laboratory 196, 197, 198, 200, 201, 205, trevally (Caranx melampygus), (USBF), 5, 71 221, 279, 280 298 US Department of Agriculture see also 1.5, 1.103–4, 1.104, Tridacna gigas (giant clam), 327 (USDA), 42 1.158 triggerfish (Balistidae), 306, 344 US Fish Commission, 66, 70 see also 1.251 walleye pollock (Theragra trophic cascades, 328 value of catch chalcogramma), 19, 241 response to fishing, 342, 344–5; fishing models, 278 wandering albatross (Diomedea tropical reef, 344 mathematical models, 280–1 exulans), 23, 359 see also 1.262, 1.277–8, 1.313, overfishing susceptibility, 330– weekend closures, 70, 71 1.313 1 weight data, 189, 279 trophic levels value of fishing effort, 254, 255 weirs, 30, 384 concept, 217–19 maximization goals, 255–6 whale fisheries, 79–80, 353 Ecopath modelling, 217 Venezuela canoe fisheries, 27 historical aspects, 69–70, 71, 75; estimation from length data, vessel monitoring systems (VMS), sustainable yield, 76, 77 206 97, 98 optimal exploitation, 259 modelling approaches, 217–19, virtual population analysis (VPA), whale products trading ban, 49 286; fractional trophic levels, 127–62, 165, 166, 278, 285 whelk (Buccinium undatum), 218–19; omnivory index, 219; basic concepts, 128–35 207 variance calculation, 219 cohort analysis, 128–30, 135 white-spotted rabbitfish (Siganus tracking food-web changes, 219 convergent estimates, 130 canaliculatus), 202, 203 see also 1.303 divergent estimates, 130 whiting see Merlangius merlangus tropical management problems, extended survivors method (XSA), wholesale markets, 51 287 149–56 wide gape hook, 20 tunas, 335 limitations, 161 winches, 14 dolphin-friendly communication multispecies, 156–61, 184, 185, World Conservation Union listing campaigns, 49 277, 285, 286 criteria, 333 fisheries bycatch, 359; dolphins, separable, 130, 135, 135–41, 136; World Health Organization, 88 17–18, 360 interpretation of results, 139, World Meteorological Organization, long line fishing, 21, 22, 23 140, 140, 141; practical 88 migration: individual-based implementation, 138–9 World War-related fishing reduction models, 241–2; see also 1.176, short-term forecasting (‘Great Fishing Experiments’), 1.180 applications, 166, 167, 171 68, 68, 75, 301 purse seining, 16–17 trial analysis, 130–1, 131, 132, schooling behaviour modelling, 135; judicious averaging Xiphias gladius (swordfish), 21, 319, 242 method (JAM), 131–5, 132, 320, 335 see also 1.72, 1.75, 1.192, 1.207, 133 communication campaigns, 49, 1.251, 1.267, 1.268, 1.342 use of additional data (tuning), 54 turbot, 177 141–9, 142; abundance indices, long lining, 21, 22 turtles, 29, 306, 328, 335 145–6; ad hoc methods, 145–6; see also 1.251 bycatch, 360 catch per unit effort (CPUE), habitat destruction, 360 141–5, 144, 146, 148; year-class strength tusk (Brosme brosme), 21, 24 catchability analysis, 141–5, dynamic pool short-term 144; data inclusion/exclusion, forecasting, 166 Ulanowicz’s theory of ascendancy, 146–7; fitting/suppressing estimation methods, 166 220–1 trends, 148–9; Hybrid method, variability, 72, 72; oceanographic underwater visual surveys, marine 146, 148; Laurec–Shepherd variables, 72–3; see also 1.137 protected areas, 294, 296, method, 144, 146, 149; Mixed virtual population analysis (VPA), 302–3 method, 146; oldest age group F 130, 132, 133, 133, 134; United Nations Agreement on values, 147–8; pseudo-code for separable, 139, 140, 141 Straddling Stocks and Highly ad hoc tuning, 146; statistical yellowfin tuna (Thunnus albacares), Migratory Stocks, 267 diagnostics, 148 195, 241 United Nations Convention on the virtual population, 128–30, 129 long lining, 21 Law of the Sea (UNCLOS), 84 visual senses see also 1.213–15, 1.214 410 Index yellowstrip goat fish marine protected areas (MPAs), multispecies estimation, 204–5, (Mulloides flavolineatus), 300–1, 303, 304 205, 213–14 298 yield-per-recruit, 128, 173, 272, 278, stock–recruitment relationship, yield 287 173 biomass relationship, 106 forecasting, 75, 76, 286 fishing models, 278 marine protected areas (MPAs), Zambia, 30 fishing rate trade-off, 73, 74 301, 303 Zoarces (eelpout), 300